A model is still running

A model is still running or downloading. Leaving will stop the current operation. Stay on this page to let it finish.

Skip to content
LocalAI.toolsBeta

Join the waitlist

Get an email when the new apps and model runs launch.

No account needed.

LAION CLAP

Compare a recording against your text descriptions and return complete ranked sound-label scores using LAION CLAP. Audio is processed locally in deterministic10second windows.

Desktop inferencePrivate · Runs on your computer
About this model

Compare a recording against your text descriptions and return complete ranked sound-label scores using LAION CLAP. Audio is processed locally in deterministic10second windows.

Created by
LAION
License
Apache-2.0
Model released
Unknown
Source repository created
Feb 16, 2023
Model weights · FP32
614.5 MB

Best for

  • Sorting authorized sound clips with user-provided descriptions
  • Exploring sound/text similarity and relative sound-label matches

Keep in mind

  • Scores are relative to the supplied labels, not calibrated event probabilities or proof that a sound occurred. A poor label set can force a misleading best match.
  • Silence and unknown sounds still receive labels; there is no verified unknown/rejection threshold.
  • English synthetic speech/tonal music/beep/noise fixtures are checked; broad natural sounds, multilingual descriptions, overlapping events and safety use are unverified.
  • Long audio is averaged over deterministic10second window logits; brief events can be diluted. No event timestamps or source separation are returned.
  • Stereo is mixed to mono, audio is resampled to48kHz and short windows use upstream repeat padding. Full input/output/IPC and processing time grow with duration and label count.
  • Text labels must fit this checkpoint’s native token context; overly long labels fail validation rather than silently truncating.
  • Only FP32 CPU and Metal on the recorded M3 Max/macOS host are verified. Windows, Intel Mac, other chips/OS versions and precisions remain unverified.
  • The app downloads isolated app-owned Python/framework packages on request; user Python and global packages are not used.

Processed on your deviceYour input is processed locally in the desktop app. Model and runtime files download from external hosts.

Target platforms · verification required

  • macos · aarch64, x86_64
  • windows · x86_64

Recommended configuration

FP32 · mps (verified platform only)

Sources & technical details
Original model
https://huggingface.co/laion/clap-htsat-unfused
Release verification
Unknown
Release evidence
Exact original checkpoint public release date has not been verified. Repository creation, training, paper, artifact conversion and update dates are not release evidence.
Source updated
Apr 24, 2023
Native model files
https://huggingface.co/laion/clap-htsat-unfused
Files updated
Apr 24, 2023
Artifact revision
8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a

Desktop app

Sample result

LAION CLAP — Sound Classification

Run LAION CLAP on your computer

Download the desktop app to load this model, provide input and view results locally.

Public desktop downloads are coming soon.

Device requirements

macOS Apple SiliconCPU / FP32Tested8.0 GiBRAM estimate
macOS Apple SiliconMPS / FP32Tested8.0 GiBRAM estimate
Windows Intel / AMDCPU / FP32Unverified8.0 GiBRAM estimate
macOS Intel / AMDCPU / FP32Unverified8.0 GiBRAM estimate
If your device falls short

MemoryAdvisory only: close other apps, reduce image/tile/token/batch settings or choose a smaller verified model/precision. CPU can need more RAM than GPU.

GPUIf the selected GPU runtime fails, existing CPU fallback may recover when enabled, but CPU has its own memory/runtime needs.

PlatformUse a compatible OS/architecture/runtime or another model. These estimates do not add hardware detection or block runs.

Benchmark evidence

Saved desktop measurements apply to the recorded hardware and runtime. Fallback cases report the device that actually ran; they do not prove GPU support on the requested device.

LAION CLAP — Sound Classification

60 configurations tested
Passed60
Failed0
TestedNov 2, 2026
Runtimetransformers 5.22.1

Quality criteria

{ "version": 1, "requirements": [ "Exact original labels returned once each", "Finite normalized sorted probabilities", "Expected sound top1 on authored semantic fixtures", "Silence structural only: no calibrated absence/rejection claim", "CPU/Metal score MAE<=.001", "Long audio deterministic10second window mean similarity, not random truncation" ], "scope": "Authored synthetic speech/tonal music/beep/noise and silence/stereo/long robustness; not broad real sound-event accuracy" }

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms0.012 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6434 ms / 0 ms0.013 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6434 ms / 0 ms0.013 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6435 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6436 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6446 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6446 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6440 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6442 ms / 0 ms0.017 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6443 ms / 0 ms0.017 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6445 ms / 0 ms0.018 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64220 ms / 0 ms0.018 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6450 ms / 0 ms0.020 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6471 ms / 0 ms0.028 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms0.037 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64274 ms / 447 ms0.107 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64283 ms / 451 ms0.111 RTF
FP32mpsAcceleratedPassedSlowmacos · aarch642689 ms / 349 ms1.050 RTF
FP32cpuPassedVery fastmacos · aarch64103 ms / 0 ms0.009 RTF
FP32cpuPassedVery fastmacos · aarch64118 ms / 0 ms0.010 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.020 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.021 RTF
FP32cpuPassedVery fastmacos · aarch6464 ms / 0 ms0.021 RTF
FP32cpuPassedVery fastmacos · aarch6460 ms / 0 ms0.023 RTF
FP32cpuPassedVery fastmacos · aarch6460 ms / 0 ms0.023 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6471 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6471 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6472 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6463 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6463 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6463 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6463 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6465 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6466 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6466 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6468 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6469 ms / 0 ms0.027 RTF
FP32cpuPassedVery fastmacos · aarch6470 ms / 0 ms0.027 RTF
FP32cpuPassedVery fastmacos · aarch6470 ms / 0 ms0.027 RTF
FP32cpuPassedVery fastmacos · aarch6479 ms / 0 ms0.031 RTF
FP32cpuPassedVery fastmacos · aarch6486 ms / 171 ms0.034 RTF
FP32cpuPassedVery fastmacos · aarch6487 ms / 177 ms0.034 RTF
FP32cpuPassedVery fastmacos · aarch6493 ms / 166 ms0.036 RTF
FP32cpuPassedVery fastmacos · aarch6498 ms / 176 ms0.038 RTF
FP32cpuPassedVery fastmacos · aarch64100 ms / 182 ms0.039 RTF
FP32cpuPassedVery fastmacos · aarch64115 ms / 205 ms0.045 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.061 RTF
FP32cpuPassedVery fastmacos · aarch6470 ms / 0 ms0.070 RTF
Test setup & sample details
Quality gate
{ "version": 1, "requirements": [ "Exact original labels returned once each", "Finite normalized sorted probabilities", "Expected sound top1 on authored semantic fixtures", "Silence structural only: no calibrated absence/rejection claim", "CPU/Metal score MAE<=.001", "Long audio deterministic10second window mean similarity, not random truncation" ], "scope": "Authored synthetic speech/tonal music/beep/noise and silence/stereo/long robustness; not broad real sound-event accuracy" }
Tested
2026-11-02T00:24:28.214336+00:00
Environment
{ "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2" }
Runtime
transformers 5.22.1
Artifact revision
8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a
Fixture
{
  "file": "model-samples/desktop/clap/clap-speech.wav",
  "url": "https://assets.localai.tools/samples/v1/files/db23763d70d55295b93980bbbc9e165f663cd456776ebcdc3ca402a6b55d7b6f/clap-speech.wav",
  "size": 491600,
  "sha1": "6c8a55b7a6660f6b640e4b2d900b9c38841b2d39",
  "sha256": "db23763d70d55295b93980bbbc9e165f663cd456776ebcdc3ca402a6b55d7b6f"
}
Rerun when
Artifact/processor/tokenizer revision; Pinned runtime version; Fixture/label checksum or quality gate; Window/aggregation/resample behavior; Relevant backend or OS execution behavior
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 205.16520900127944,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 114.97099999542115,
      "inferenceMs": 114.97099999542115,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.04491054687321139,
      "audioSecondsPerSecond": 22.266484592653406,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 320.1362089967006,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 71.37154200609075,
      "inferenceMs": 71.37154200609075,
      "audioSeconds": 3,
      "realTimeFactor": 0.02379051400203025,
      "audioSecondsPerSecond": 42.0335600951985,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 71.37154200609075,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 71.42683299753116,
      "inferenceMs": 71.42683299753116,
      "audioSeconds": 3,
      "realTimeFactor": 0.02380894433251039,
      "audioSecondsPerSecond": 42.001022222330555,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 71.42683299753116,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 71.94316699315095,
      "inferenceMs": 71.94316699315095,
      "audioSeconds": 3,
      "realTimeFactor": 0.02398105566438365,
      "audioSecondsPerSecond": 41.69958211994757,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 71.94316699315095,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 70.21070799964946,
      "inferenceMs": 70.21070799964946,
      "audioSeconds": 1,
      "realTimeFactor": 0.07021070799964946,
      "audioSecondsPerSecond": 14.242841704501723,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 70.21070799964946,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 117.98458299745107,
      "inferenceMs": 117.98458299745107,
      "audioSeconds": 12,
      "realTimeFactor": 0.009832048583120923,
      "audioSecondsPerSecond": 101.70820369182681,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 117.98458299745107,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 69.45029200142017,
      "inferenceMs": 69.45029200142017,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.027129020313054752,
      "audioSecondsPerSecond": 36.860896134859324,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 69.45029200142017,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 69.90187499468448,
      "inferenceMs": 69.90187499468448,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.027305419919798624,
      "audioSecondsPerSecond": 36.622765844187576,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 69.90187499468448,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 69.51529200159712,
      "inferenceMs": 69.51529200159712,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.027154410938123874,
      "audioSecondsPerSecond": 36.8264294990113,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 69.51529200159712,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 165.77624999627005,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 92.61862499988638,
      "inferenceMs": 92.61862499988638,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03617915039058062,
      "audioSecondsPerSecond": 27.64022894966472,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 258.39487499615643,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 66.81629100057762,
      "inferenceMs": 66.81629100057762,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026100113672100633,
      "audioSecondsPerSecond": 38.31400937801036,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 66.81629100057762,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.62308299948927,
      "inferenceMs": 67.62308299948927,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026415266796675496,
      "audioSecondsPerSecond": 37.85689569964349,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 67.62308299948927,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 182.33999999938533,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 99.89220799616305,
      "inferenceMs": 99.89220799616305,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03902039374850119,
      "audioSecondsPerSecond": 25.627624530016714,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 282.2322079955484,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 78.86420800059568,
      "inferenceMs": 78.86420800059568,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.030806331250232688,
      "audioSecondsPerSecond": 32.460859810836666,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 78.86420800059568,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.28999999904772,
      "inferenceMs": 67.28999999904772,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026285156249628017,
      "audioSecondsPerSecond": 38.04428592712481,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 67.28999999904772,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 66.85825000022305,
      "inferenceMs": 66.85825000022305,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02611650390633713,
      "audioSecondsPerSecond": 38.289964215208435,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 66.85825000022305,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 66.51966600475134,
      "inferenceMs": 66.51966600475134,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.025984244533105993,
      "audioSecondsPerSecond": 38.484859497297315,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 66.51966600475134,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.2340000019176,
      "inferenceMs": 67.2340000019176,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02626328125074906,
      "audioSecondsPerSecond": 38.07597346472002,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 67.2340000019176,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 65.95083299907856,
      "inferenceMs": 65.95083299907856,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02576204414026506,
      "audioSecondsPerSecond": 38.816795536695764,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 65.95083299907856,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 65.55787500110455,
      "inferenceMs": 65.55787500110455,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.025608544922306464,
      "audioSecondsPerSecond": 39.04946583391954,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 65.55787500110455,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 170.82125000160886,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 86.34512500429992,
      "inferenceMs": 86.34512500429992,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.033728564454804655,
      "audioSecondsPerSecond": 29.648460175053476,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 257.1663750059088,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 63.71075000060955,
      "inferenceMs": 63.71075000060955,
      "audioSeconds": 3,
      "realTimeFactor": 0.02123691666686985,
      "audioSecondsPerSecond": 47.087814850261495,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 63.71075000060955,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.03875000210246,
      "inferenceMs": 62.03875000210246,
      "audioSeconds": 3,
      "realTimeFactor": 0.020679583334034152,
      "audioSecondsPerSecond": 48.35687372647469,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.03875000210246,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.28708300093422,
      "inferenceMs": 61.28708300093422,
      "audioSeconds": 3,
      "realTimeFactor": 0.020429027666978072,
      "audioSecondsPerSecond": 48.9499557346247,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.28708300093422,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 60.92754199926276,
      "inferenceMs": 60.92754199926276,
      "audioSeconds": 1,
      "realTimeFactor": 0.06092754199926276,
      "audioSecondsPerSecond": 16.412938503445623,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 60.92754199926276,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 103.38729099748889,
      "inferenceMs": 103.38729099748889,
      "audioSeconds": 12,
      "realTimeFactor": 0.008615607583124074,
      "audioSecondsPerSecond": 116.06842469923562,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 103.38729099748889,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.825208002119325,
      "inferenceMs": 61.825208002119325,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02415047187582786,
      "audioSecondsPerSecond": 41.40705842691617,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.825208002119325,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.26458299352089,
      "inferenceMs": 62.26458299352089,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.024322102731844097,
      "audioSecondsPerSecond": 41.11486621963545,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.26458299352089,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 60.722249996615574,
      "inferenceMs": 60.722249996615574,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02371962890492796,
      "audioSecondsPerSecond": 42.15917559284586,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 60.722249996615574,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 176.01504200138152,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 98.27249999943888,
      "inferenceMs": 98.27249999943888,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03838769531228081,
      "audioSecondsPerSecond": 26.050013991855476,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 274.2875420008204,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 65.06616700062295,
      "inferenceMs": 65.06616700062295,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02541647148461834,
      "audioSecondsPerSecond": 39.34456443354793,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 65.06616700062295,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 63.25133399514016,
      "inferenceMs": 63.25133399514016,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.024707552341851624,
      "audioSecondsPerSecond": 40.47345468155177,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 63.25133399514016,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 177.4701250033104,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 86.57879100064747,
      "inferenceMs": 86.57879100064747,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03381984023462792,
      "audioSecondsPerSecond": 29.568442460473435,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 264.04891600395786,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.89204199856613,
      "inferenceMs": 62.89204199856613,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.024567203905689894,
      "audioSecondsPerSecond": 40.70467293872196,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.89204199856613,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.444082999718376,
      "inferenceMs": 62.444082999718376,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02439221992176499,
      "audioSecondsPerSecond": 40.9966785806038,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.444082999718376,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.47337500099093,
      "inferenceMs": 62.47337500099093,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02440366210976208,
      "audioSecondsPerSecond": 40.97745639577491,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.47337500099093,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.607541993202176,
      "inferenceMs": 62.607541993202176,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.0244560710910946,
      "audioSecondsPerSecond": 40.889642341779854,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.607541993202176,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 59.859957997105084,
      "inferenceMs": 59.859957997105084,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.023382796092619174,
      "audioSecondsPerSecond": 42.76648507043399,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 59.859957997105084,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.56249999569263,
      "inferenceMs": 62.56249999569263,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.024438476560817435,
      "audioSecondsPerSecond": 40.91908092189816,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.56249999569263,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 60.15950000437442,
      "inferenceMs": 60.15950000437442,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02349980468920876,
      "audioSecondsPerSecond": 42.55354515602445,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 60.15950000437442,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 349.2796670034295,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 2688.5440420010127,
      "inferenceMs": 2688.5440420010127,
      "audioSeconds": 2.56,
      "realTimeFactor": 1.0502125164066456,
      "audioSecondsPerSecond": 0.9521882327412644,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3037.823709004442,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "slow"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.818422555923462e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 45.5172920046607,
      "inferenceMs": 45.5172920046607,
      "audioSeconds": 3,
      "realTimeFactor": 0.015172430668220235,
      "audioSecondsPerSecond": 65.90901760352565,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 45.5172920046607,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.221794031560421e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 46.21483400114812,
      "inferenceMs": 46.21483400114812,
      "audioSeconds": 3,
      "realTimeFactor": 0.015404944667049373,
      "audioSecondsPerSecond": 64.91422212888335,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 46.21483400114812,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 6.859863788122311e-10
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 37.413833997561596,
      "inferenceMs": 37.413833997561596,
      "audioSeconds": 3,
      "realTimeFactor": 0.0124712779991872,
      "audioSecondsPerSecond": 80.18424415406135,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 37.413833997561596,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 8.23289155960083e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 37.348916004702915,
      "inferenceMs": 37.348916004702915,
      "audioSeconds": 1,
      "realTimeFactor": 0.037348916004702915,
      "audioSecondsPerSecond": 26.774538781100947,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 37.348916004702915,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.2281816452741623e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 220.41041599732125,
      "inferenceMs": 220.41041599732125,
      "audioSeconds": 12,
      "realTimeFactor": 0.018367534666443436,
      "audioSecondsPerSecond": 54.44388798824209,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 220.41041599732125,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.3493051887489855e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 50.476208998588845,
      "inferenceMs": 50.476208998588845,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.019717269140073768,
      "audioSecondsPerSecond": 50.71696252132504,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 50.476208998588845,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 42.86458300339291,
      "inferenceMs": 42.86458300339291,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.016743977735700355,
      "audioSecondsPerSecond": 59.72296522276597,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 42.86458300339291,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 33.98920899780933,
      "inferenceMs": 33.98920899780933,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.013277034764769269,
      "audioSecondsPerSecond": 75.3180222630364,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 33.98920899780933,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 447.2832909959834,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 273.7911250005709,
      "inferenceMs": 273.7911250005709,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.10694965820334801,
      "audioSecondsPerSecond": 9.350193509722647,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 721.0744159965543,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 39.57529200124554,
      "inferenceMs": 39.57529200124554,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01545909843798654,
      "audioSecondsPerSecond": 64.68682530300546,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 39.57529200124554,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 36.98729200550588,
      "inferenceMs": 36.98729200550588,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.014448160939650734,
      "audioSecondsPerSecond": 69.21296102507104,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.98729200550588,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 451.08600000094157,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 283.469125002739,
      "inferenceMs": 283.469125002739,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.11073012695419493,
      "audioSecondsPerSecond": 9.030965894346568,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 734.5551250036806,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 42.37295799975982,
      "inferenceMs": 42.37295799975982,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01655193671865618,
      "audioSecondsPerSecond": 60.41589072007932,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 42.37295799975982,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 36.3219589999062,
      "inferenceMs": 36.3219589999062,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01418826523433836,
      "audioSecondsPerSecond": 70.48077996031577,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.3219589999062,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 34.42120800173143,
      "inferenceMs": 34.42120800173143,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01344578437567634,
      "audioSecondsPerSecond": 74.37275298040757,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 34.42120800173143,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 70.77741700049955,
      "inferenceMs": 70.77741700049955,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02764742851582014,
      "audioSecondsPerSecond": 36.169729109808166,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 70.77741700049955,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 37.45133300253656,
      "inferenceMs": 37.45133300253656,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.014629426954115843,
      "audioSecondsPerSecond": 68.35537735937497,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 37.45133300253656,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 44.99404200032586,
      "inferenceMs": 44.99404200032586,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01757579765637729,
      "audioSecondsPerSecond": 56.89642197474634,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 44.99404200032586,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.57160684466362e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 34.85050000017509,
      "inferenceMs": 34.85050000017509,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.013613476562568394,
      "audioSecondsPerSecond": 73.45662185584536,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 34.85050000017509,
      "artifactRevision": "8fa0f1c6d0433df6e97c127f64b2a1d6c0dcda8a",
      "fallback": false,
      "grade": "fast"
    }
  }
]

LAION CLAP 630k Fused — Sound Classification

60 configurations tested
Passed60
Failed0
TestedNov 5, 2026
Runtimetransformers 5.22.1

Quality criteria

{ "version": 1, "requirements": [ "Exact original labels returned once each", "Finite normalized sorted probabilities", "Expected sound top1 on authored semantic fixtures", "Silence structural only: no calibrated absence/rejection claim", "CPU/Metal score MAE<=.001", "Long audio deterministic10second window mean similarity, not random truncation" ], "scope": "Authored synthetic speech/tonal music/beep/noise and silence/stereo/long robustness; not broad real sound-event accuracy" }

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch6472 ms / 0 ms0.006 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms0.012 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6436 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6443 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6438 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6439 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6446 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6439 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6439 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6439 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6442 ms / 0 ms0.016 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6442 ms / 0 ms0.017 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6444 ms / 0 ms0.017 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6447 ms / 0 ms0.018 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6448 ms / 0 ms0.019 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6452 ms / 0 ms0.020 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6447 ms / 0 ms0.047 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64324 ms / 3434 ms0.126 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64332 ms / 3436 ms0.130 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64341 ms / 3559 ms0.133 RTF
FP32cpuPassedVery fastmacos · aarch64106 ms / 0 ms0.009 RTF
FP32cpuPassedVery fastmacos · aarch64128 ms / 0 ms0.011 RTF
FP32cpuPassedVery fastmacos · aarch6464 ms / 0 ms0.021 RTF
FP32cpuPassedVery fastmacos · aarch6465 ms / 0 ms0.022 RTF
FP32cpuPassedVery fastmacos · aarch6466 ms / 0 ms0.022 RTF
FP32cpuPassedVery fastmacos · aarch6459 ms / 0 ms0.023 RTF
FP32cpuPassedVery fastmacos · aarch6460 ms / 0 ms0.023 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6474 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6463 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6463 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6475 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6465 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6465 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6476 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6466 ms / 3238 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6466 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 3211 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6468 ms / 3362 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6474 ms / 0 ms0.029 RTF
FP32cpuPassedVery fastmacos · aarch6475 ms / 0 ms0.029 RTF
FP32cpuPassedVery fastmacos · aarch6475 ms / 0 ms0.029 RTF
FP32cpuPassedVery fastmacos · aarch6475 ms / 0 ms0.029 RTF
FP32cpuPassedVery fastmacos · aarch6475 ms / 0 ms0.029 RTF
FP32cpuPassedVery fastmacos · aarch6475 ms / 0 ms0.029 RTF
FP32cpuPassedVery fastmacos · aarch6475 ms / 0 ms0.029 RTF
FP32cpuPassedVery fastmacos · aarch6476 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6476 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6476 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6476 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6478 ms / 3257 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6479 ms / 3320 ms0.031 RTF
FP32cpuPassedVery fastmacos · aarch6481 ms / 0 ms0.032 RTF
FP32cpuPassedVery fastmacos · aarch6482 ms / 3287 ms0.032 RTF
FP32cpuPassedVery fastmacos · aarch6465 ms / 0 ms0.065 RTF
FP32cpuPassedVery fastmacos · aarch6474 ms / 0 ms0.074 RTF
Test setup & sample details
Quality gate
{ "version": 1, "requirements": [ "Exact original labels returned once each", "Finite normalized sorted probabilities", "Expected sound top1 on authored semantic fixtures", "Silence structural only: no calibrated absence/rejection claim", "CPU/Metal score MAE<=.001", "Long audio deterministic10second window mean similarity, not random truncation" ], "scope": "Authored synthetic speech/tonal music/beep/noise and silence/stereo/long robustness; not broad real sound-event accuracy" }
Tested
2026-11-05T19:30:21.310271+00:00
Environment
{ "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2" }
Runtime
transformers 5.22.1
Artifact revision
b3708341862f581175dba5c356a4ebf74a9b6651
Fixture
{
  "file": "model-samples/desktop/clap/clap-speech.wav",
  "url": "https://assets.localai.tools/samples/v1/files/db23763d70d55295b93980bbbc9e165f663cd456776ebcdc3ca402a6b55d7b6f/clap-speech.wav",
  "size": 491600,
  "sha1": "6c8a55b7a6660f6b640e4b2d900b9c38841b2d39",
  "sha256": "db23763d70d55295b93980bbbc9e165f663cd456776ebcdc3ca402a6b55d7b6f"
}
Rerun when
Artifact/processor/tokenizer revision; Pinned runtime version; Fixture/label checksum or quality gate; Window/aggregation/resample behavior; Relevant backend or OS execution behavior
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3286.9860000209883,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 81.91691699903458,
      "inferenceMs": 81.91691699903458,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.031998795702747884,
      "audioSecondsPerSecond": 31.251176115797552,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 3368.902917020023,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 76.36895799078047,
      "inferenceMs": 76.36895799078047,
      "audioSeconds": 3,
      "realTimeFactor": 0.025456319330260158,
      "audioSecondsPerSecond": 39.282976734633074,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 76.36895799078047,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 75.18362504197285,
      "inferenceMs": 75.18362504197285,
      "audioSeconds": 3,
      "realTimeFactor": 0.025061208347324282,
      "audioSecondsPerSecond": 39.90230583222326,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 75.18362504197285,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 73.64112499635667,
      "inferenceMs": 73.64112499635667,
      "audioSeconds": 3,
      "realTimeFactor": 0.024547041665452223,
      "audioSecondsPerSecond": 40.738106596666235,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 73.64112499635667,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 73.8367079757154,
      "inferenceMs": 73.8367079757154,
      "audioSeconds": 1,
      "realTimeFactor": 0.0738367079757154,
      "audioSecondsPerSecond": 13.543399041150318,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 73.8367079757154,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 127.98670900519937,
      "inferenceMs": 127.98670900519937,
      "audioSeconds": 12,
      "realTimeFactor": 0.010665559083766615,
      "audioSecondsPerSecond": 93.75973562623997,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 127.98670900519937,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 75.6794170010835,
      "inferenceMs": 75.6794170010835,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02956227226604824,
      "audioSecondsPerSecond": 33.82689906244057,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 75.6794170010835,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 75.09770797332749,
      "inferenceMs": 75.09770797332749,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02933504217708105,
      "audioSecondsPerSecond": 34.08892320534253,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 75.09770797332749,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 76.30854198941961,
      "inferenceMs": 76.30854198941961,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.029808024214617035,
      "audioSecondsPerSecond": 33.5480135415895,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 76.30854198941961,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3257.3690410354175,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 77.94316700892523,
      "inferenceMs": 77.94316700892523,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.030446549612861418,
      "audioSecondsPerSecond": 32.84444420518422,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 3335.3122080443427,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 75.4640829982236,
      "inferenceMs": 75.4640829982236,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.029478157421181095,
      "audioSecondsPerSecond": 33.92342288264818,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 75.4640829982236,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 75.9891250054352,
      "inferenceMs": 75.9891250054352,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.029683251955248124,
      "audioSecondsPerSecond": 33.68903115829921,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 75.9891250054352,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3320.379375014454,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 79.30549996672198,
      "inferenceMs": 79.30549996672198,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.030978710924500774,
      "audioSecondsPerSecond": 32.28023278428636,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 3399.684874981176,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 81.1360829975456,
      "inferenceMs": 81.1360829975456,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03169378242091625,
      "audioSecondsPerSecond": 31.551929861803178,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 81.1360829975456,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 76.38412498636171,
      "inferenceMs": 76.38412498636171,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.029837548822797544,
      "audioSecondsPerSecond": 33.5148173845951,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 76.38412498636171,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 74.57312499172986,
      "inferenceMs": 74.57312499172986,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.029130126949894475,
      "audioSecondsPerSecond": 34.32872097399571,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 74.57312499172986,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 75.15337504446507,
      "inferenceMs": 75.15337504446507,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.029356787126744166,
      "audioSecondsPerSecond": 34.06367310164522,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 75.15337504446507,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 74.56483400892466,
      "inferenceMs": 74.56483400892466,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.029126888284736197,
      "audioSecondsPerSecond": 34.332538039226286,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 74.56483400892466,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 73.71045800391585,
      "inferenceMs": 73.71045800391585,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.028793147657779627,
      "audioSecondsPerSecond": 34.73048559627727,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 73.71045800391585,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 75.38412499707192,
      "inferenceMs": 75.38412499707192,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02944692382698122,
      "audioSecondsPerSecond": 33.9594045841805,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 75.38412499707192,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3361.738834006246,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.83320900285617,
      "inferenceMs": 67.83320900285617,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02649734726674069,
      "audioSecondsPerSecond": 37.7396269118303,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3429.5720430091023,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 66.33091700496152,
      "inferenceMs": 66.33091700496152,
      "audioSeconds": 3,
      "realTimeFactor": 0.022110305668320507,
      "audioSecondsPerSecond": 45.22777816829521,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 66.33091700496152,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 65.11775002581999,
      "inferenceMs": 65.11775002581999,
      "audioSeconds": 3,
      "realTimeFactor": 0.02170591667527333,
      "audioSecondsPerSecond": 46.070387856006434,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 65.11775002581999,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 64.25341701833531,
      "inferenceMs": 64.25341701833531,
      "audioSeconds": 3,
      "realTimeFactor": 0.02141780567277844,
      "audioSecondsPerSecond": 46.6901238753407,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 64.25341701833531,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 64.5867909770459,
      "inferenceMs": 64.5867909770459,
      "audioSeconds": 1,
      "realTimeFactor": 0.0645867909770459,
      "audioSecondsPerSecond": 15.48304204114119,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 64.5867909770459,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 105.62645801110193,
      "inferenceMs": 105.62645801110193,
      "audioSeconds": 12,
      "realTimeFactor": 0.008802204834258495,
      "audioSecondsPerSecond": 113.60789925132899,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 105.62645801110193,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 63.38083400623873,
      "inferenceMs": 63.38083400623873,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.024758138283687003,
      "audioSecondsPerSecond": 40.39075913308451,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 63.38083400623873,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 66.07975001679733,
      "inferenceMs": 66.07975001679733,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02581240235031146,
      "audioSecondsPerSecond": 38.74106665580989,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 66.07975001679733,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 65.08337502600625,
      "inferenceMs": 65.08337502600625,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.025423193369533692,
      "audioSecondsPerSecond": 39.33416174218171,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 65.08337502600625,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3210.8415830298327,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.09183298517019,
      "inferenceMs": 67.09183298517019,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026207747259832104,
      "audioSecondsPerSecond": 38.15665612483499,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3277.933416015003,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 59.38454100396484,
      "inferenceMs": 59.38454100396484,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.023197086329673766,
      "audioSecondsPerSecond": 43.1088622850361,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 59.38454100396484,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 64.75666601909325,
      "inferenceMs": 64.75666601909325,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.0252955726637083,
      "audioSecondsPerSecond": 39.53260965049056,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 64.75666601909325,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3237.640167004429,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 65.66120800562203,
      "inferenceMs": 65.66120800562203,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.025648909377196105,
      "audioSecondsPerSecond": 38.98801252302285,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3303.301375010051,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.421666003298014,
      "inferenceMs": 61.421666003298014,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.023992838282538287,
      "audioSecondsPerSecond": 41.67910391526244,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.421666003298014,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.18245801189914,
      "inferenceMs": 61.18245801189914,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.023899397660898103,
      "audioSecondsPerSecond": 41.84205870745035,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.18245801189914,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.45724996551871,
      "inferenceMs": 62.45724996551871,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.024397363267780747,
      "audioSecondsPerSecond": 40.988035839127086,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.45724996551871,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.93133303895593,
      "inferenceMs": 61.93133303895593,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02419192696834216,
      "audioSecondsPerSecond": 41.33610362285781,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.93133303895593,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 59.53008303185925,
      "inferenceMs": 59.53008303185925,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02325393868432002,
      "audioSecondsPerSecond": 43.00346765231189,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 59.53008303185925,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 63.47929104231298,
      "inferenceMs": 63.47929104231298,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.024796598063403508,
      "audioSecondsPerSecond": 40.32811264847929,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 63.47929104231298,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.26070801494643,
      "inferenceMs": 62.26070801494643,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02432058906833845,
      "audioSecondsPerSecond": 41.11742512445315,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.26070801494643,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3434.249208017718,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 323.6022919882089,
      "inferenceMs": 323.6022919882089,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.1264071453078941,
      "audioSecondsPerSecond": 7.9109452045947775,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3757.851500005927,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.2351741790771484e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 37.289583997335285,
      "inferenceMs": 37.289583997335285,
      "audioSeconds": 3,
      "realTimeFactor": 0.012429861332445094,
      "audioSecondsPerSecond": 80.45142043457444,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 37.289583997335285,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.2992026060819626e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 45.94391694990918,
      "inferenceMs": 45.94391694990918,
      "audioSeconds": 3,
      "realTimeFactor": 0.01531463898330306,
      "audioSecondsPerSecond": 65.29700119540918,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 45.94391694990918,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 8.492406777804717e-11
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 43.14487497322261,
      "inferenceMs": 43.14487497322261,
      "audioSeconds": 3,
      "realTimeFactor": 0.014381624991074204,
      "audioSecondsPerSecond": 69.5331717118642,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 43.14487497322261,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 7.129274308681488e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 46.926833980251104,
      "inferenceMs": 46.926833980251104,
      "audioSeconds": 1,
      "realTimeFactor": 0.046926833980251104,
      "audioSecondsPerSecond": 21.30976917004127,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 46.926833980251104,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.766035661101341e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 71.78929197834805,
      "inferenceMs": 71.78929197834805,
      "audioSeconds": 12,
      "realTimeFactor": 0.0059824409981956705,
      "audioSecondsPerSecond": 167.15584830700448,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 71.78929197834805,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.01865690946579e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 47.02395899221301,
      "inferenceMs": 47.02395899221301,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.018368733981333207,
      "audioSecondsPerSecond": 54.440333286781026,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 47.02395899221301,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 41.59870895091444,
      "inferenceMs": 41.59870895091444,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.016249495683950954,
      "audioSecondsPerSecond": 61.5403714336602,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 41.59870895091444,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 51.734042004682124,
      "inferenceMs": 51.734042004682124,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.020208610158078955,
      "audioSecondsPerSecond": 49.48385822565943,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 51.734042004682124,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3558.630124956835,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 340.89483303250745,
      "inferenceMs": 340.89483303250745,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.13316204415332322,
      "audioSecondsPerSecond": 7.5096474101028114,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3899.5249579893425,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 39.41816696897149,
      "inferenceMs": 39.41816696897149,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.015397721472254489,
      "audioSecondsPerSecond": 64.94467391178125,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 39.41816696897149,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 36.49204096291214,
      "inferenceMs": 36.49204096291214,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.014254703501137556,
      "audioSecondsPerSecond": 70.15228341439708,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.49204096291214,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3435.924875026103,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 331.97241701418534,
      "inferenceMs": 331.97241701418534,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.12967672539616615,
      "audioSecondsPerSecond": 7.7114840534796905,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3767.8972920402884,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 44.199957977980375,
      "inferenceMs": 44.199957977980375,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.017265608585148584,
      "audioSecondsPerSecond": 57.91860710083358,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 44.199957977980375,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 42.41891700075939,
      "inferenceMs": 42.41891700075939,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.016569889453421638,
      "audioSecondsPerSecond": 60.35043280228419,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 42.41891700075939,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 39.3204580177553,
      "inferenceMs": 39.3204580177553,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.015359553913185664,
      "audioSecondsPerSecond": 65.10605748396986,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 39.3204580177553,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 38.530000019818544,
      "inferenceMs": 38.530000019818544,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.015050781257741619,
      "audioSecondsPerSecond": 66.4417336798137,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.530000019818544,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 38.3892499958165,
      "inferenceMs": 38.3892499958165,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01499580077961582,
      "audioSecondsPerSecond": 66.68533509456367,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.3892499958165,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 39.26829097326845,
      "inferenceMs": 39.26829097326845,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.015339176161432988,
      "audioSecondsPerSecond": 65.1925494221966,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 39.26829097326845,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 4.353933036327362e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 47.88966604974121,
      "inferenceMs": 47.88966604974121,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01870690080068016,
      "audioSecondsPerSecond": 53.456209056480446,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 47.88966604974121,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-fused-native"
    }
  }
]

LAION CLAP 630k Unfused — Sound Classification

60 configurations tested
Passed60
Failed0
TestedNov 5, 2026
Runtimetransformers 5.22.1

Quality criteria

{ "version": 1, "requirements": [ "Exact original labels returned once each", "Finite normalized sorted probabilities", "Expected sound top1 on authored semantic fixtures", "Silence structural only: no calibrated absence/rejection claim", "CPU/Metal score MAE<=.001", "Long audio deterministic10second window mean similarity, not random truncation" ], "scope": "Authored synthetic speech/tonal music/beep/noise and silence/stereo/long robustness; not broad real sound-event accuracy" }

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch6465 ms / 0 ms0.005 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6435 ms / 0 ms0.012 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6438 ms / 0 ms0.013 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6440 ms / 0 ms0.013 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6434 ms / 0 ms0.013 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6435 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6435 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6435 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6435 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6436 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6436 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6436 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6438 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms0.037 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64286 ms / 3538 ms0.112 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64294 ms / 3509 ms0.115 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64305 ms / 3627 ms0.119 RTF
FP32cpuPassedVery fastmacos · aarch64100 ms / 0 ms0.008 RTF
FP32cpuPassedVery fastmacos · aarch64113 ms / 0 ms0.009 RTF
FP32cpuPassedVery fastmacos · aarch6459 ms / 0 ms0.020 RTF
FP32cpuPassedVery fastmacos · aarch6460 ms / 0 ms0.020 RTF
FP32cpuPassedVery fastmacos · aarch6464 ms / 0 ms0.021 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.022 RTF
FP32cpuPassedVery fastmacos · aarch6468 ms / 0 ms0.023 RTF
FP32cpuPassedVery fastmacos · aarch6468 ms / 0 ms0.023 RTF
FP32cpuPassedVery fastmacos · aarch6460 ms / 0 ms0.023 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6463 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6463 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6466 ms / 3516 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6466 ms / 3386 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6468 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6468 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6468 ms / 0 ms0.027 RTF
FP32cpuPassedVery fastmacos · aarch6468 ms / 0 ms0.027 RTF
FP32cpuPassedVery fastmacos · aarch6468 ms / 0 ms0.027 RTF
FP32cpuPassedVery fastmacos · aarch6468 ms / 0 ms0.027 RTF
FP32cpuPassedVery fastmacos · aarch6469 ms / 0 ms0.027 RTF
FP32cpuPassedVery fastmacos · aarch6469 ms / 0 ms0.027 RTF
FP32cpuPassedVery fastmacos · aarch6470 ms / 0 ms0.027 RTF
FP32cpuPassedVery fastmacos · aarch6471 ms / 3257 ms0.028 RTF
FP32cpuPassedVery fastmacos · aarch6472 ms / 3220 ms0.028 RTF
FP32cpuPassedVery fastmacos · aarch6472 ms / 3249 ms0.028 RTF
FP32cpuPassedVery fastmacos · aarch6472 ms / 3284 ms0.028 RTF
FP32cpuPassedVery fastmacos · aarch6476 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6479 ms / 0 ms0.031 RTF
FP32cpuPassedVery fastmacos · aarch6458 ms / 0 ms0.058 RTF
FP32cpuPassedVery fastmacos · aarch6467 ms / 0 ms0.067 RTF
Test setup & sample details
Quality gate
{ "version": 1, "requirements": [ "Exact original labels returned once each", "Finite normalized sorted probabilities", "Expected sound top1 on authored semantic fixtures", "Silence structural only: no calibrated absence/rejection claim", "CPU/Metal score MAE<=.001", "Long audio deterministic10second window mean similarity, not random truncation" ], "scope": "Authored synthetic speech/tonal music/beep/noise and silence/stereo/long robustness; not broad real sound-event accuracy" }
Tested
2026-11-05T19:29:21.490813+00:00
Environment
{ "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2" }
Runtime
transformers 5.22.1
Artifact revision
b3708341862f581175dba5c356a4ebf74a9b6651
Fixture
{
  "file": "model-samples/desktop/clap/clap-speech.wav",
  "url": "https://assets.localai.tools/samples/v1/files/db23763d70d55295b93980bbbc9e165f663cd456776ebcdc3ca402a6b55d7b6f/clap-speech.wav",
  "size": 491600,
  "sha1": "6c8a55b7a6660f6b640e4b2d900b9c38841b2d39",
  "sha256": "db23763d70d55295b93980bbbc9e165f663cd456776ebcdc3ca402a6b55d7b6f"
}
Rerun when
Artifact/processor/tokenizer revision; Pinned runtime version; Fixture/label checksum or quality gate; Window/aggregation/resample behavior; Relevant backend or OS execution behavior
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3257.2215420077555,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 71.37929199961945,
      "inferenceMs": 71.37929199961945,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02788253593735135,
      "audioSecondsPerSecond": 35.86474351712046,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 3328.600834007375,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.59441702160984,
      "inferenceMs": 67.59441702160984,
      "audioSeconds": 3,
      "realTimeFactor": 0.022531472340536613,
      "audioSecondsPerSecond": 44.382363694932145,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 67.59441702160984,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 68.39937495533377,
      "inferenceMs": 68.39937495533377,
      "audioSeconds": 3,
      "realTimeFactor": 0.022799791651777923,
      "audioSecondsPerSecond": 43.86004991944829,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 68.39937495533377,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 66.53358298353851,
      "inferenceMs": 66.53358298353851,
      "audioSeconds": 3,
      "realTimeFactor": 0.022177860994512837,
      "audioSecondsPerSecond": 45.09001117138467,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 66.53358298353851,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.16149998828769,
      "inferenceMs": 67.16149998828769,
      "audioSeconds": 1,
      "realTimeFactor": 0.06716149998828769,
      "audioSecondsPerSecond": 14.889482816411043,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 67.16149998828769,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 112.63304203748703,
      "inferenceMs": 112.63304203748703,
      "audioSeconds": 12,
      "realTimeFactor": 0.009386086836457253,
      "audioSecondsPerSecond": 106.54067210584712,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 112.63304203748703,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 70.2537500183098,
      "inferenceMs": 70.2537500183098,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.027442871100902266,
      "audioSecondsPerSecond": 36.43933596900953,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 70.2537500183098,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.72608298342675,
      "inferenceMs": 67.72608298342675,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026455501165401074,
      "audioSecondsPerSecond": 37.799321726999295,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 67.72608298342675,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 68.39500000933185,
      "inferenceMs": 68.39500000933185,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026716796878645255,
      "audioSecondsPerSecond": 37.429636664240256,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 68.39500000933185,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3283.6252080160193,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 72.01962498947978,
      "inferenceMs": 72.01962498947978,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02813266601151554,
      "audioSecondsPerSecond": 35.54586684357146,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 3355.644833005499,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 68.18924995604903,
      "inferenceMs": 68.18924995604903,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02663642576408165,
      "audioSecondsPerSecond": 37.54257455024117,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 68.18924995604903,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 66.81841699173674,
      "inferenceMs": 66.81841699173674,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026100944137397164,
      "audioSecondsPerSecond": 38.312790324209395,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 66.81841699173674,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3249.0059579722583,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 71.69520802563056,
      "inferenceMs": 71.69520802563056,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02800594063501194,
      "audioSecondsPerSecond": 35.706709981018776,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 3320.701165997889,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 78.60258297296241,
      "inferenceMs": 78.60258297296241,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03070413397381344,
      "audioSecondsPerSecond": 32.56890426718655,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 78.60258297296241,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 76.40641700709239,
      "inferenceMs": 76.40641700709239,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.029846256643395463,
      "audioSecondsPerSecond": 33.50503923986344,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 76.40641700709239,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 68.60145798418671,
      "inferenceMs": 68.60145798418671,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026797444525072933,
      "audioSecondsPerSecond": 37.31699114310522,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 68.60145798418671,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.27466703159735,
      "inferenceMs": 67.27466703159735,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026279166809217713,
      "audioSecondsPerSecond": 38.05295682545151,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 67.27466703159735,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 66.83987501310185,
      "inferenceMs": 66.83987501310185,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02610932617699291,
      "audioSecondsPerSecond": 38.3004905305133,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 66.83987501310185,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.15070799691603,
      "inferenceMs": 67.15070799691603,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026230745311295323,
      "audioSecondsPerSecond": 38.12320191944322,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 67.15070799691603,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.72233301308006,
      "inferenceMs": 67.72233301308006,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.0264540363332344,
      "audioSecondsPerSecond": 37.80141477857172,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 67.72233301308006,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3516.2259999779053,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 65.57233299827203,
      "inferenceMs": 65.57233299827203,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.025614192577450012,
      "audioSecondsPerSecond": 39.0408558449104,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3581.7983329761773,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 59.58412500331178,
      "inferenceMs": 59.58412500331178,
      "audioSeconds": 3,
      "realTimeFactor": 0.019861375001103927,
      "audioSecondsPerSecond": 50.34898137437204,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 59.58412500331178,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 63.906958035659045,
      "inferenceMs": 63.906958035659045,
      "audioSeconds": 3,
      "realTimeFactor": 0.021302319345219683,
      "audioSecondsPerSecond": 46.94324518350645,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 63.906958035659045,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 58.57908300822601,
      "inferenceMs": 58.57908300822601,
      "audioSeconds": 3,
      "realTimeFactor": 0.019526361002742004,
      "audioSecondsPerSecond": 51.21281942188687,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 58.57908300822601,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 58.23854199843481,
      "inferenceMs": 58.23854199843481,
      "audioSeconds": 1,
      "realTimeFactor": 0.05823854199843481,
      "audioSecondsPerSecond": 17.17075952943457,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 58.23854199843481,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 99.88191694719717,
      "inferenceMs": 99.88191694719717,
      "audioSeconds": 12,
      "realTimeFactor": 0.008323493078933097,
      "audioSecondsPerSecond": 120.14186718446574,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 99.88191694719717,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.07391702244058,
      "inferenceMs": 61.07391702244058,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.023856998836890853,
      "audioSecondsPerSecond": 41.916420704756355,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.07391702244058,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 59.89591695833951,
      "inferenceMs": 59.89591695833951,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.023396842561851372,
      "audioSecondsPerSecond": 42.74080989160919,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 59.89591695833951,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 60.95008295960724,
      "inferenceMs": 60.95008295960724,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02380862615609658,
      "audioSecondsPerSecond": 42.001583520346635,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 60.95008295960724,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3219.743166002445,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 71.64220802951604,
      "inferenceMs": 71.64220802951604,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.027985237511529704,
      "audioSecondsPerSecond": 35.733125351821926,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3291.385374031961,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 63.22608300251886,
      "inferenceMs": 63.22608300251886,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02469768867285893,
      "audioSecondsPerSecond": 40.489618815987896,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 63.22608300251886,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.837041983380914,
      "inferenceMs": 61.837041983380914,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02415509452475817,
      "audioSecondsPerSecond": 41.39913420645211,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.837041983380914,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3386.1290420172736,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 65.832499996759,
      "inferenceMs": 65.832499996759,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.025715820311233983,
      "audioSecondsPerSecond": 38.886568186321824,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3451.9615420140326,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.295708001125604,
      "inferenceMs": 61.295708001125604,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02394363593793969,
      "audioSecondsPerSecond": 41.76475129307568,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.295708001125604,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 63.42725001741201,
      "inferenceMs": 63.42725001741201,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.024776269538051565,
      "audioSecondsPerSecond": 40.361201207639155,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 63.42725001741201,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.94337497558445,
      "inferenceMs": 67.94337497558445,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026540380849837675,
      "audioSecondsPerSecond": 37.678434445152895,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 67.94337497558445,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.2741670277901,
      "inferenceMs": 67.2741670277901,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026278971495230508,
      "audioSecondsPerSecond": 38.05323964758266,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 67.2741670277901,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 69.36266703996807,
      "inferenceMs": 69.36266703996807,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02709479181248753,
      "audioSecondsPerSecond": 36.90746202888767,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 69.36266703996807,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 66.50491699110717,
      "inferenceMs": 66.50491699110717,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.025978483199651237,
      "audioSecondsPerSecond": 38.4933944108571,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 66.50491699110717,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 68.22970899520442,
      "inferenceMs": 68.22970899520442,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026652230076251726,
      "audioSecondsPerSecond": 37.52031245186656,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 68.22970899520442,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3538.011832977645,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 286.10691596986726,
      "inferenceMs": 286.10691596986726,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.1117605140507294,
      "audioSecondsPerSecond": 8.947704012403596,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3824.1187489475124,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.7462298274040222e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 39.94554199744016,
      "inferenceMs": 39.94554199744016,
      "audioSeconds": 3,
      "realTimeFactor": 0.013315180665813386,
      "audioSecondsPerSecond": 75.10224796029178,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 39.94554199744016,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 5.727633833885193e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 37.55399998044595,
      "inferenceMs": 37.55399998044595,
      "audioSeconds": 3,
      "realTimeFactor": 0.012517999993481984,
      "audioSecondsPerSecond": 79.88496569106019,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 37.55399998044595,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.9243096832942683e-10
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 35.45300004770979,
      "inferenceMs": 35.45300004770979,
      "audioSeconds": 3,
      "realTimeFactor": 0.01181766668256993,
      "audioSecondsPerSecond": 84.61907302521202,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.45300004770979,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0.0000015422701835632324
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 36.697582982014865,
      "inferenceMs": 36.697582982014865,
      "audioSeconds": 1,
      "realTimeFactor": 0.036697582982014865,
      "audioSecondsPerSecond": 27.24975103919216,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.697582982014865,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.562308847904205e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 65.10091602103785,
      "inferenceMs": 65.10091602103785,
      "audioSeconds": 12,
      "realTimeFactor": 0.005425076335086487,
      "audioSecondsPerSecond": 184.32920354180746,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 65.10091602103785,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 8.86757334228605e-10
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 36.269792006351054,
      "inferenceMs": 36.269792006351054,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01416788750248088,
      "audioSecondsPerSecond": 70.58215276094577,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.269792006351054,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 34.68849998898804,
      "inferenceMs": 34.68849998898804,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.013550195308198454,
      "audioSecondsPerSecond": 73.79967426705332,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 34.68849998898804,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 34.48320901952684,
      "inferenceMs": 34.48320901952684,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.013470003523252672,
      "audioSecondsPerSecond": 74.23903032198501,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 34.48320901952684,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3626.5569999814034,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 304.9880829639733,
      "inferenceMs": 304.9880829639733,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.11913596990780206,
      "audioSecondsPerSecond": 8.393770586447472,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3931.5450829453766,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 36.228790995664895,
      "inferenceMs": 36.228790995664895,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.0141518714826816,
      "audioSecondsPerSecond": 70.66203231309396,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.228790995664895,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 35.349750018212944,
      "inferenceMs": 35.349750018212944,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.013808496100864431,
      "audioSecondsPerSecond": 72.41918255945328,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.349750018212944,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 3509.0963330003433,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 294.17916701640934,
      "inferenceMs": 294.17916701640934,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.1149137371157849,
      "audioSecondsPerSecond": 8.702179783713927,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 3803.2755000167526,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 36.59558395156637,
      "inferenceMs": 36.59558395156637,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.014295149981080613,
      "audioSecondsPerSecond": 69.95379561064298,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.59558395156637,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 36.84316697763279,
      "inferenceMs": 36.84316697763279,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.014391862100637809,
      "audioSecondsPerSecond": 69.48371190658384,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.84316697763279,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 36.65904200170189,
      "inferenceMs": 36.65904200170189,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.014319938281914801,
      "audioSecondsPerSecond": 69.83270320815127,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135493120,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 36.65904200170189,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 37.6334169995971,
      "inferenceMs": 37.6334169995971,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.014700553515467618,
      "audioSecondsPerSecond": 68.02464947648541,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 37.6334169995971,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 34.967957995831966,
      "inferenceMs": 34.967957995831966,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.013659358592121862,
      "audioSecondsPerSecond": 73.20987975063174,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 34.967957995831966,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 35.530833003576845,
      "inferenceMs": 35.530833003576845,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.013879231642022205,
      "audioSecondsPerSecond": 72.05009800198853,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.530833003576845,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.0199877326376736e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 430
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 35.437332990113646,
      "inferenceMs": 35.437332990113646,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.013842708199263143,
      "audioSecondsPerSecond": 72.24019936021122,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1135476736,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 35.437332990113646,
      "artifactRevision": "b3708341862f581175dba5c356a4ebf74a9b6651",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-630k-unfused-native"
    }
  }
]

LAION CLAP HTSAT Fused — Sound Classification

60 configurations tested
Passed60
Failed0
TestedNov 5, 2026
Runtimetransformers 5.22.1

Quality criteria

{ "version": 1, "requirements": [ "Exact original labels returned once each", "Finite normalized sorted probabilities", "Expected sound top1 on authored semantic fixtures", "Silence structural only: no calibrated absence/rejection claim", "CPU/Metal score MAE<=.001", "Long audio deterministic10second window mean similarity, not random truncation" ], "scope": "Authored synthetic speech/tonal music/beep/noise and silence/stereo/long robustness; not broad real sound-event accuracy" }

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch6485 ms / 0 ms0.007 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6439 ms / 0 ms0.013 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6437 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6438 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6438 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6438 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6439 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6439 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6439 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6447 ms / 0 ms0.016 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6440 ms / 0 ms0.016 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6440 ms / 0 ms0.016 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6441 ms / 0 ms0.016 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6448 ms / 0 ms0.016 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6444 ms / 0 ms0.017 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6444 ms / 0 ms0.017 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6441 ms / 0 ms0.041 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64344 ms / 484 ms0.134 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64404 ms / 579 ms0.158 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64889 ms / 455 ms0.347 RTF
FP32cpuPassedVery fastmacos · aarch64112 ms / 0 ms0.009 RTF
FP32cpuPassedVery fastmacos · aarch64128 ms / 0 ms0.011 RTF
FP32cpuPassedVery fastmacos · aarch6471 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6471 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6472 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6461 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6473 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6462 ms / 0 ms0.024 RTF
FP32cpuPassedVery fastmacos · aarch6463 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6474 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6475 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6465 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6465 ms / 0 ms0.025 RTF
FP32cpuPassedVery fastmacos · aarch6468 ms / 0 ms0.026 RTF
FP32cpuPassedVery fastmacos · aarch6469 ms / 0 ms0.027 RTF
FP32cpuPassedVery fastmacos · aarch6473 ms / 0 ms0.028 RTF
FP32cpuPassedVery fastmacos · aarch6474 ms / 0 ms0.029 RTF
FP32cpuPassedVery fastmacos · aarch6474 ms / 0 ms0.029 RTF
FP32cpuPassedVery fastmacos · aarch6475 ms / 0 ms0.029 RTF
FP32cpuPassedVery fastmacos · aarch6476 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6476 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6477 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6477 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6477 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6477 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6478 ms / 0 ms0.030 RTF
FP32cpuPassedVery fastmacos · aarch6483 ms / 0 ms0.033 RTF
FP32cpuPassedVery fastmacos · aarch6488 ms / 174 ms0.034 RTF
FP32cpuPassedVery fastmacos · aarch6497 ms / 187 ms0.038 RTF
FP32cpuPassedVery fastmacos · aarch6499 ms / 182 ms0.039 RTF
FP32cpuPassedVery fastmacos · aarch64108 ms / 188 ms0.042 RTF
FP32cpuPassedVery fastmacos · aarch64123 ms / 209 ms0.048 RTF
FP32cpuPassedVery fastmacos · aarch64125 ms / 212 ms0.049 RTF
FP32cpuPassedVery fastmacos · aarch6473 ms / 0 ms0.073 RTF
FP32cpuPassedVery fastmacos · aarch6473 ms / 0 ms0.073 RTF
Test setup & sample details
Quality gate
{ "version": 1, "requirements": [ "Exact original labels returned once each", "Finite normalized sorted probabilities", "Expected sound top1 on authored semantic fixtures", "Silence structural only: no calibrated absence/rejection claim", "CPU/Metal score MAE<=.001", "Long audio deterministic10second window mean similarity, not random truncation" ], "scope": "Authored synthetic speech/tonal music/beep/noise and silence/stereo/long robustness; not broad real sound-event accuracy" }
Tested
2026-11-05T19:20:19.036994+00:00
Environment
{ "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2" }
Runtime
transformers 5.22.1
Artifact revision
365dea6ef167def6676140ed93bbc43f84dabb28
Fixture
{
  "file": "model-samples/desktop/clap/clap-speech.wav",
  "url": "https://assets.localai.tools/samples/v1/files/db23763d70d55295b93980bbbc9e165f663cd456776ebcdc3ca402a6b55d7b6f/clap-speech.wav",
  "size": 491600,
  "sha1": "6c8a55b7a6660f6b640e4b2d900b9c38841b2d39",
  "sha256": "db23763d70d55295b93980bbbc9e165f663cd456776ebcdc3ca402a6b55d7b6f"
}
Rerun when
Artifact/processor/tokenizer revision; Pinned runtime version; Fixture/label checksum or quality gate; Window/aggregation/resample behavior; Relevant backend or OS execution behavior
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 211.95187501143664,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 124.7671659803018,
      "inferenceMs": 124.7671659803018,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.04873717421105539,
      "audioSecondsPerSecond": 20.51821871472317,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 336.71904099173844,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 74.22966696321964,
      "inferenceMs": 74.22966696321964,
      "audioSeconds": 3,
      "realTimeFactor": 0.024743222321073215,
      "audioSecondsPerSecond": 40.41510790404707,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 74.22966696321964,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 72.80145800905302,
      "inferenceMs": 72.80145800905302,
      "audioSeconds": 3,
      "realTimeFactor": 0.02426715266968434,
      "audioSecondsPerSecond": 41.207965912261585,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 72.80145800905302,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 75.28037496376783,
      "inferenceMs": 75.28037496376783,
      "audioSeconds": 3,
      "realTimeFactor": 0.02509345832125594,
      "audioSecondsPerSecond": 39.85102360932566,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 75.28037496376783,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 73.11691698851064,
      "inferenceMs": 73.11691698851064,
      "audioSeconds": 1,
      "realTimeFactor": 0.07311691698851064,
      "audioSecondsPerSecond": 13.676725458174568,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 73.11691698851064,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 128.40362504357472,
      "inferenceMs": 128.40362504357472,
      "audioSeconds": 12,
      "realTimeFactor": 0.01070030208696456,
      "audioSecondsPerSecond": 93.45530545518253,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 128.40362504357472,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 76.05129200965166,
      "inferenceMs": 76.05129200965166,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02970753594127018,
      "audioSecondsPerSecond": 33.66149255787937,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 76.05129200965166,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 75.18945902120322,
      "inferenceMs": 75.18945902120322,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.029370882430157508,
      "audioSecondsPerSecond": 34.0473256933274,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 75.18945902120322,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 72.9278750368394,
      "inferenceMs": 72.9278750368394,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02848745118626539,
      "audioSecondsPerSecond": 35.10317555128022,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 72.9278750368394,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 181.72945798141882,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 99.4267089990899,
      "inferenceMs": 99.4267089990899,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03883855820276949,
      "audioSecondsPerSecond": 25.747608723762877,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 281.1561669805087,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 74.05437499983236,
      "inferenceMs": 74.05437499983236,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.028927490234309516,
      "audioSecondsPerSecond": 34.5691932449068,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 74.05437499983236,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 73.99004098260775,
      "inferenceMs": 73.99004098260775,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.028902359758831153,
      "audioSecondsPerSecond": 34.59925100733163,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 73.99004098260775,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 188.1927500362508,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 107.51304100267589,
      "inferenceMs": 107.51304100267589,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.04199728164167027,
      "audioSecondsPerSecond": 23.811064928730683,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 295.7057910389267,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 77.68425001995638,
      "inferenceMs": 77.68425001995638,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03034541016404546,
      "audioSecondsPerSecond": 32.953912785955445,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 77.68425001995638,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 76.95391704328358,
      "inferenceMs": 76.95391704328358,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03006012384503265,
      "audioSecondsPerSecond": 33.26666267761392,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 76.95391704328358,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 77.08066696068272,
      "inferenceMs": 77.08066696068272,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.030109635531516687,
      "audioSecondsPerSecond": 33.21195963841107,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 77.08066696068272,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 83.2677079597488,
      "inferenceMs": 83.2677079597488,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03252644842177688,
      "audioSecondsPerSecond": 30.74421120415001,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 83.2677079597488,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 75.64812497003004,
      "inferenceMs": 75.64812497003004,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.029550048816417984,
      "audioSecondsPerSecond": 33.84089164158676,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 75.64812497003004,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 76.93433400709182,
      "inferenceMs": 76.93433400709182,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.030052474221520242,
      "audioSecondsPerSecond": 33.27513044779225,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 76.93433400709182,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 77.15145801194012,
      "inferenceMs": 77.15145801194012,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03013728828591411,
      "audioSecondsPerSecond": 33.181485690183706,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 77.15145801194012,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 209.2239169869572,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 123.1934170355089,
      "inferenceMs": 123.1934170355089,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.048122428529495664,
      "audioSecondsPerSecond": 20.780331137840857,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 332.4173340224661,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 71.34024996776134,
      "inferenceMs": 71.34024996776134,
      "audioSeconds": 3,
      "realTimeFactor": 0.023780083322587114,
      "audioSecondsPerSecond": 42.05199731365814,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 71.34024996776134,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 71.52233301894739,
      "inferenceMs": 71.52233301894739,
      "audioSeconds": 3,
      "realTimeFactor": 0.023840777672982465,
      "audioSecondsPerSecond": 41.94494045944017,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 71.52233301894739,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 70.68008300848305,
      "inferenceMs": 70.68008300848305,
      "audioSeconds": 3,
      "realTimeFactor": 0.02356002766949435,
      "audioSecondsPerSecond": 42.44477188347301,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 70.68008300848305,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 73.08287499472499,
      "inferenceMs": 73.08287499472499,
      "audioSeconds": 1,
      "realTimeFactor": 0.07308287499472499,
      "audioSecondsPerSecond": 13.683096075136323,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 73.08287499472499,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 112.37491702195257,
      "inferenceMs": 112.37491702195257,
      "audioSeconds": 12,
      "realTimeFactor": 0.009364576418496048,
      "audioSecondsPerSecond": 106.78539586957636,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 112.37491702195257,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 68.96824995055795,
      "inferenceMs": 68.96824995055795,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026940722636936698,
      "audioSecondsPerSecond": 37.11852920489089,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 68.96824995055795,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 64.63945802533999,
      "inferenceMs": 64.63945802533999,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.025249788291148434,
      "audioSecondsPerSecond": 39.60429245858509,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 64.63945802533999,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.27637502504513,
      "inferenceMs": 61.27637502504513,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.023936083994158253,
      "audioSecondsPerSecond": 41.77792826278752,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.27637502504513,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 187.21891700988635,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 97.43204194819555,
      "inferenceMs": 97.43204194819555,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.038059391386013885,
      "audioSecondsPerSecond": 26.27472388766262,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 284.6509589580819,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 67.63629202032462,
      "inferenceMs": 67.63629202032462,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.026420426570439304,
      "audioSecondsPerSecond": 37.84950244213156,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 67.63629202032462,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 64.71941701602191,
      "inferenceMs": 64.71941701602191,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.025281022271883558,
      "audioSecondsPerSecond": 39.55536248675182,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 64.71941701602191,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 174.18087500846013,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 87.96524995705113,
      "inferenceMs": 87.96524995705113,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.0343614257644731,
      "audioSecondsPerSecond": 29.102401246513995,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 262.14612496551126,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.80570896342397,
      "inferenceMs": 62.80570896342397,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.024533480063837487,
      "audioSecondsPerSecond": 40.760625781501204,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.80570896342397,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.12508400110528,
      "inferenceMs": 61.12508400110528,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02387698593793175,
      "audioSecondsPerSecond": 41.881333037574386,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.12508400110528,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.21308304136619,
      "inferenceMs": 61.21308304136619,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.023911360563033668,
      "audioSecondsPerSecond": 41.821125040704445,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.21308304136619,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.09395797830075,
      "inferenceMs": 61.09395797830075,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02386482733527373,
      "audioSecondsPerSecond": 41.902670652133175,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.09395797830075,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.203167006373405,
      "inferenceMs": 62.203167006373405,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02429811211186461,
      "audioSecondsPerSecond": 41.15546077802919,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.203167006373405,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 62.24270898383111,
      "inferenceMs": 62.24270898383111,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.024313558196809026,
      "audioSecondsPerSecond": 41.12931525305262,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 62.24270898383111,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 61.98733299970627,
      "inferenceMs": 61.98733299970627,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.02421380195301026,
      "audioSecondsPerSecond": 41.29876018399002,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 61.98733299970627,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 455.24979097535834,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 888.7434159987606,
      "inferenceMs": 888.7434159987606,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.34716539687451586,
      "audioSecondsPerSecond": 2.8804714093134502,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 1343.993206974119,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 6.555637810379267e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 46.58366699004546,
      "inferenceMs": 46.58366699004546,
      "audioSeconds": 3,
      "realTimeFactor": 0.015527888996681819,
      "audioSecondsPerSecond": 64.40025429172579,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 46.58366699004546,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.3684257282875478e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 48.12304099323228,
      "inferenceMs": 48.12304099323228,
      "audioSeconds": 3,
      "realTimeFactor": 0.01604101366441076,
      "audioSecondsPerSecond": 62.3401999973755,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 48.12304099323228,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 7.350564601438236e-12
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 39.27599999587983,
      "inferenceMs": 39.27599999587983,
      "audioSeconds": 3,
      "realTimeFactor": 0.01309199999862661,
      "audioSecondsPerSecond": 76.38252368659509,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 39.27599999587983,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0.0000015918631106615067
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 41.21183301322162,
      "inferenceMs": 41.21183301322162,
      "audioSeconds": 1,
      "realTimeFactor": 0.04121183301322162,
      "audioSecondsPerSecond": 24.264875568120907,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 41.21183301322162,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.1641532182693481e-8
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 85.36420803284273,
      "inferenceMs": 85.36420803284273,
      "audioSeconds": 12,
      "realTimeFactor": 0.0071136840027368935,
      "audioSecondsPerSecond": 140.57413846542292,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 85.36420803284273,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.7017009668052197e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 37.79841598588973,
      "inferenceMs": 37.79841598588973,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.014765006244488177,
      "audioSecondsPerSecond": 67.7277058635382,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 37.79841598588973,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 37.303792021702975,
      "inferenceMs": 37.303792021702975,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.014571793758477725,
      "audioSecondsPerSecond": 68.62573109218005,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 37.303792021702975,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 40.96283297985792,
      "inferenceMs": 40.96283297985792,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.016001106632757,
      "audioSecondsPerSecond": 62.495677514755705,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 40.96283297985792,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 484.10241602687165,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 343.81291700992733,
      "inferenceMs": 343.81291700992733,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.13430192070700286,
      "audioSecondsPerSecond": 7.445909892693421,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 827.915333036799,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 38.41320803621784,
      "inferenceMs": 38.41320803621784,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.015005159389147593,
      "audioSecondsPerSecond": 66.6437439327199,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.41320803621784,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 38.63249998539686,
      "inferenceMs": 38.63249998539686,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01509082030679565,
      "audioSecondsPerSecond": 66.26545009946764,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.63249998539686,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 579.0005840244703,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 404.03833403252065,
      "inferenceMs": 404.03833403252065,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.15782747423145338,
      "audioSecondsPerSecond": 6.336032461201932,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 983.038918056991,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 40.10237497277558,
      "inferenceMs": 40.10237497277558,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01566499022374046,
      "audioSecondsPerSecond": 63.83661819874546,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136754688,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 40.10237497277558,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 39.223416009917855,
      "inferenceMs": 39.223416009917855,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.015321646878874162,
      "audioSecondsPerSecond": 65.26713530898711,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 39.223416009917855,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 43.98483398836106,
      "inferenceMs": 43.98483398836106,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01718157577670354,
      "audioSecondsPerSecond": 58.20187932680178,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 43.98483398836106,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 38.29083399614319,
      "inferenceMs": 38.29083399614319,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.014957357029743434,
      "audioSecondsPerSecond": 66.85673130697162,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.29083399614319,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 38.57770899776369,
      "inferenceMs": 38.57770899776369,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.015069417577251443,
      "audioSecondsPerSecond": 66.35956531655108,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 38.57770899776369,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 40.052999975159764,
      "inferenceMs": 40.052999975159764,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.015645703115296783,
      "audioSecondsPerSecond": 63.91531225095927,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 40.052999975159764,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.166060196235776e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 448
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 43.961624964140356,
      "inferenceMs": 43.961624964140356,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.017172509751617326,
      "audioSecondsPerSecond": 58.23260632627208,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1136738304,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 43.961624964140356,
      "artifactRevision": "365dea6ef167def6676140ed93bbc43f84dabb28",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-htsat-fused-native"
    }
  }
]

LAION CLAP Larger General — Sound Classification

60 configurations tested
Passed60
Failed0
TestedNov 5, 2026
Runtimetransformers 5.22.1

Quality criteria

{ "version": 1, "requirements": [ "Exact original labels returned once each", "Finite normalized sorted probabilities", "Expected sound top1 on authored semantic fixtures", "Silence structural only: no calibrated absence/rejection claim", "CPU/Metal score MAE<=.001", "Long audio deterministic10second window mean similarity, not random truncation" ], "scope": "Authored synthetic speech/tonal music/beep/noise and silence/stereo/long robustness; not broad real sound-event accuracy" }

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch6486 ms / 0 ms0.007 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6441 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6442 ms / 0 ms0.014 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6444 ms / 0 ms0.015 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6441 ms / 0 ms0.016 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6442 ms / 0 ms0.017 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6443 ms / 0 ms0.017 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6443 ms / 0 ms0.017 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6443 ms / 0 ms0.017 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6444 ms / 0 ms0.017 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6446 ms / 0 ms0.018 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6446 ms / 0 ms0.018 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6446 ms / 0 ms0.018 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6446 ms / 0 ms0.018 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6447 ms / 0 ms0.018 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6447 ms / 0 ms0.018 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch6442 ms / 0 ms0.042 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64274 ms / 588 ms0.107 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64283 ms / 569 ms0.111 RTF
FP32mpsAcceleratedPassedFastmacos · aarch64385 ms / 422 ms0.150 RTF
FP32cpuPassedVery fastmacos · aarch64154 ms / 0 ms0.013 RTF
FP32cpuPassedVery fastmacos · aarch64175 ms / 0 ms0.015 RTF
FP32cpuPassedVery fastmacos · aarch6483 ms / 0 ms0.028 RTF
FP32cpuPassedVery fastmacos · aarch6484 ms / 0 ms0.028 RTF
FP32cpuPassedVery fastmacos · aarch6486 ms / 0 ms0.029 RTF
FP32cpuPassedVery fastmacos · aarch6481 ms / 0 ms0.031 RTF
FP32cpuPassedVery fastmacos · aarch6481 ms / 0 ms0.032 RTF
FP32cpuPassedVery fastmacos · aarch6481 ms / 0 ms0.032 RTF
FP32cpuPassedVery fastmacos · aarch6481 ms / 0 ms0.032 RTF
FP32cpuPassedVery fastmacos · aarch6481 ms / 0 ms0.032 RTF
FP32cpuPassedVery fastmacos · aarch6481 ms / 0 ms0.032 RTF
FP32cpuPassedVery fastmacos · aarch6482 ms / 0 ms0.032 RTF
FP32cpuPassedVery fastmacos · aarch6485 ms / 0 ms0.033 RTF
FP32cpuPassedVery fastmacos · aarch6485 ms / 0 ms0.033 RTF
FP32cpuPassedVery fastmacos · aarch64100 ms / 0 ms0.033 RTF
FP32cpuPassedVery fastmacos · aarch64100 ms / 0 ms0.033 RTF
FP32cpuPassedVery fastmacos · aarch6486 ms / 0 ms0.033 RTF
FP32cpuPassedVery fastmacos · aarch6487 ms / 0 ms0.034 RTF
FP32cpuPassedVery fastmacos · aarch6488 ms / 0 ms0.034 RTF
FP32cpuPassedVery fastmacos · aarch64103 ms / 0 ms0.034 RTF
FP32cpuPassedVery fastmacos · aarch6498 ms / 0 ms0.038 RTF
FP32cpuPassedVery fastmacos · aarch6498 ms / 0 ms0.038 RTF
FP32cpuPassedVery fastmacos · aarch6498 ms / 0 ms0.038 RTF
FP32cpuPassedVery fastmacos · aarch6498 ms / 0 ms0.038 RTF
FP32cpuPassedVery fastmacos · aarch6499 ms / 0 ms0.039 RTF
FP32cpuPassedVery fastmacos · aarch6499 ms / 0 ms0.039 RTF
FP32cpuPassedVery fastmacos · aarch6499 ms / 0 ms0.039 RTF
FP32cpuPassedVery fastmacos · aarch6499 ms / 0 ms0.039 RTF
FP32cpuPassedVery fastmacos · aarch6499 ms / 0 ms0.039 RTF
FP32cpuPassedVery fastmacos · aarch64100 ms / 0 ms0.039 RTF
FP32cpuPassedVery fastmacos · aarch64101 ms / 0 ms0.039 RTF
FP32cpuPassedVery fastmacos · aarch64102 ms / 0 ms0.040 RTF
FP32cpuPassedVery fastmacos · aarch64120 ms / 174 ms0.047 RTF
FP32cpuPassedVery fastmacos · aarch64122 ms / 183 ms0.048 RTF
FP32cpuPassedVery fastmacos · aarch64124 ms / 178 ms0.048 RTF
FP32cpuPassedVery fastmacos · aarch64134 ms / 180 ms0.052 RTF
FP32cpuPassedVery fastmacos · aarch64137 ms / 181 ms0.054 RTF
FP32cpuPassedVery fastmacos · aarch64138 ms / 186 ms0.054 RTF
FP32cpuPassedVery fastmacos · aarch6483 ms / 0 ms0.083 RTF
FP32cpuPassedFastmacos · aarch64101 ms / 0 ms0.101 RTF
Test setup & sample details
Quality gate
{ "version": 1, "requirements": [ "Exact original labels returned once each", "Finite normalized sorted probabilities", "Expected sound top1 on authored semantic fixtures", "Silence structural only: no calibrated absence/rejection claim", "CPU/Metal score MAE<=.001", "Long audio deterministic10second window mean similarity, not random truncation" ], "scope": "Authored synthetic speech/tonal music/beep/noise and silence/stereo/long robustness; not broad real sound-event accuracy" }
Tested
2026-11-05T19:21:07.469510+00:00
Environment
{ "os": "macos", "arch": "aarch64", "hardware": "Apple M3 Max", "osVersion": "27.2" }
Runtime
transformers 5.22.1
Artifact revision
ada0c23a36c4e8582805bb38fec3905903f18b41
Fixture
{
  "file": "model-samples/desktop/clap/clap-speech.wav",
  "url": "https://assets.localai.tools/samples/v1/files/db23763d70d55295b93980bbbc9e165f663cd456776ebcdc3ca402a6b55d7b6f/clap-speech.wav",
  "size": 491600,
  "sha1": "6c8a55b7a6660f6b640e4b2d900b9c38841b2d39",
  "sha256": "db23763d70d55295b93980bbbc9e165f663cd456776ebcdc3ca402a6b55d7b6f"
}
Rerun when
Artifact/processor/tokenizer revision; Pinned runtime version; Fixture/label checksum or quality gate; Window/aggregation/resample behavior; Relevant backend or OS execution behavior
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 179.86937501700595,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 134.1475830413401,
      "inferenceMs": 134.1475830413401,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.05240139962552348,
      "audioSecondsPerSecond": 19.08345973860064,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 314.01695805834606,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 102.95083303935826,
      "inferenceMs": 102.95083303935826,
      "audioSeconds": 3,
      "realTimeFactor": 0.03431694434645275,
      "audioSecondsPerSecond": 29.14012360495515,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 102.95083303935826,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 99.8398750089109,
      "inferenceMs": 99.8398750089109,
      "audioSeconds": 3,
      "realTimeFactor": 0.03327995833630363,
      "audioSecondsPerSecond": 30.04811454073079,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 99.8398750089109,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 100.14300001785159,
      "inferenceMs": 100.14300001785159,
      "audioSeconds": 3,
      "realTimeFactor": 0.03338100000595053,
      "audioSecondsPerSecond": 29.957161254058867,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 100.14300001785159,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 100.7261669728905,
      "inferenceMs": 100.7261669728905,
      "audioSeconds": 1,
      "realTimeFactor": 0.1007261669728905,
      "audioSecondsPerSecond": 9.92790681957689,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 100.7261669728905,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 175.26375001762062,
      "inferenceMs": 175.26375001762062,
      "audioSeconds": 12,
      "realTimeFactor": 0.014605312501468385,
      "audioSecondsPerSecond": 68.46823714997281,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 175.26375001762062,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 100.6687919725664,
      "inferenceMs": 100.6687919725664,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03932374686428375,
      "audioSecondsPerSecond": 25.429926691656682,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 100.6687919725664,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 98.20454195141792,
      "inferenceMs": 98.20454195141792,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.038361149199772626,
      "audioSecondsPerSecond": 26.068040735493064,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 98.20454195141792,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 98.7410829984583,
      "inferenceMs": 98.7410829984583,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03857073554627277,
      "audioSecondsPerSecond": 25.9263917536733,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 98.7410829984583,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 181.31800001719967,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 137.45787501102313,
      "inferenceMs": 137.45787501102313,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.05369448242618091,
      "audioSecondsPerSecond": 18.62388749858607,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 318.7758750282228,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 98.6980419838801,
      "inferenceMs": 98.6980419838801,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.038553922649953165,
      "audioSecondsPerSecond": 25.93769793749417,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 98.6980419838801,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 98.04720798274502,
      "inferenceMs": 98.04720798274502,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.038299690618259774,
      "audioSecondsPerSecond": 26.10987148609602,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 98.04720798274502,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 185.75858400436118,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 137.82233401434496,
      "inferenceMs": 137.82233401434496,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.0538368492243535,
      "audioSecondsPerSecond": 18.57463827113498,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 323.58091801870614,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 97.74279198609293,
      "inferenceMs": 97.74279198609293,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03818077811956755,
      "audioSecondsPerSecond": 26.191189631295195,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 97.74279198609293,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 98.34362502442673,
      "inferenceMs": 98.34362502442673,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03841547852516669,
      "audioSecondsPerSecond": 26.03117384949094,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 98.34362502442673,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 99.93054199730977,
      "inferenceMs": 99.93054199730977,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03903536796769913,
      "audioSecondsPerSecond": 25.617793607773265,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 99.93054199730977,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 99.32204202050343,
      "inferenceMs": 99.32204202050343,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03879767266425915,
      "audioSecondsPerSecond": 25.774741919538158,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 99.32204202050343,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 98.65037497365847,
      "inferenceMs": 98.65037497365847,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03853530272408534,
      "audioSecondsPerSecond": 25.950230809397013,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 98.65037497365847,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 98.89475000090897,
      "inferenceMs": 98.89475000090897,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03863076171910507,
      "audioSecondsPerSecond": 25.88610618841213,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 98.89475000090897,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 102.43554197950289,
      "inferenceMs": 102.43554197950289,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.040013883585743315,
      "audioSecondsPerSecond": 24.99132576964595,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 102.43554197950289,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 183.23374999454245,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 121.6419159900397,
      "inferenceMs": 121.6419159900397,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.04751637343360926,
      "audioSecondsPerSecond": 21.04537715609164,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 304.87566598458216,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 83.97266699466854,
      "inferenceMs": 83.97266699466854,
      "audioSeconds": 3,
      "realTimeFactor": 0.027990888998222847,
      "audioSecondsPerSecond": 35.72591067270105,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 83.97266699466854,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 82.53175002755597,
      "inferenceMs": 82.53175002755597,
      "audioSeconds": 3,
      "realTimeFactor": 0.027510583342518657,
      "audioSecondsPerSecond": 36.349647244828205,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 82.53175002755597,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 85.58958303183317,
      "inferenceMs": 85.58958303183317,
      "audioSeconds": 3,
      "realTimeFactor": 0.028529861010611057,
      "audioSecondsPerSecond": 35.05099445202596,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 85.58958303183317,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 82.99537503626198,
      "inferenceMs": 82.99537503626198,
      "audioSeconds": 1,
      "realTimeFactor": 0.08299537503626198,
      "audioSecondsPerSecond": 12.048864163371565,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 82.99537503626198,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 153.8000000291504,
      "inferenceMs": 153.8000000291504,
      "audioSeconds": 12,
      "realTimeFactor": 0.012816666669095866,
      "audioSecondsPerSecond": 78.02340700731851,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 153.8000000291504,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 87.56499999435619,
      "inferenceMs": 87.56499999435619,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.034205078122795385,
      "audioSecondsPerSecond": 29.235425114657676,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 87.56499999435619,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 87.14204200077802,
      "inferenceMs": 87.14204200077802,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.034039860156553914,
      "audioSecondsPerSecond": 29.377323978443652,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 87.14204200077802,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 85.50308301346377,
      "inferenceMs": 85.50308301346377,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03339964180213428,
      "audioSecondsPerSecond": 29.940440856347706,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 85.50308301346377,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 177.95829195529222,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 123.76375001622364,
      "inferenceMs": 123.76375001622364,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.04834521485008736,
      "audioSecondsPerSecond": 20.6845703985571,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 301.72204197151586,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 84.83637496829033,
      "inferenceMs": 84.83637496829033,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03313920897198841,
      "audioSecondsPerSecond": 30.175735360649988,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 84.83637496829033,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 85.02358297118917,
      "inferenceMs": 85.02358297118917,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03321233709812077,
      "audioSecondsPerSecond": 30.109293334150287,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 85.02358297118917,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 173.62725001294166,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 120.25849998462945,
      "inferenceMs": 120.25849998462945,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.04697597655649588,
      "audioSecondsPerSecond": 21.287476563629184,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 293.8857499975711,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 81.10666699940339,
      "inferenceMs": 81.10666699940339,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03168229179664195,
      "audioSecondsPerSecond": 31.563373206037813,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 81.10666699940339,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 80.96262498293072,
      "inferenceMs": 80.96262498293072,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03162602538395731,
      "audioSecondsPerSecond": 31.619528153141314,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 80.96262498293072,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 80.95345902256668,
      "inferenceMs": 80.95345902256668,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03162244493069011,
      "audioSecondsPerSecond": 31.6231082761562,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 80.95345902256668,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 81.5689159790054,
      "inferenceMs": 81.5689159790054,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03186285780429898,
      "audioSecondsPerSecond": 31.3845043700091,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 81.5689159790054,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 80.58470801915973,
      "inferenceMs": 80.58470801915973,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03147840156998427,
      "audioSecondsPerSecond": 31.767813806452423,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 80.58470801915973,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 80.69329202407971,
      "inferenceMs": 80.69329202407971,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03152081719690614,
      "audioSecondsPerSecond": 31.72506581136967,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 80.69329202407971,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 0
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": {
        "cpu/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 81.49758301442489,
      "inferenceMs": 81.49758301442489,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.03183499336500972,
      "audioSecondsPerSecond": 31.41197450661679,
      "activationPlacement": {
        "output": "cpu"
      },
      "scorePlacement": "cpu",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": null,
      "device": "cpu",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 81.49758301442489,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 421.8591249664314,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 385.0886669824831,
      "inferenceMs": 385.0886669824831,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.15042526054003247,
      "audioSecondsPerSecond": 6.647819630891524,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177436160,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 806.9477919489145,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a musical instrument playing a melody.",
      "scope": "top1",
      "cpuAgreementScoreMae": 5.458132363855839e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 42.461832985281944,
      "inferenceMs": 42.461832985281944,
      "audioSeconds": 3,
      "realTimeFactor": 0.014153944328427315,
      "audioSecondsPerSecond": 70.65168385547217,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177436160,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 42.461832985281944,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 3.290097083663568e-10
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 40.59300001244992,
      "inferenceMs": 40.59300001244992,
      "audioSeconds": 3,
      "realTimeFactor": 0.013531000004149973,
      "audioSecondsPerSecond": 73.90436772546735,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177436160,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 40.59300001244992,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of white noise or static.",
      "scope": "top1",
      "cpuAgreementScoreMae": 7.611689056830073e-13
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 43.56949997600168,
      "inferenceMs": 43.56949997600168,
      "audioSeconds": 3,
      "realTimeFactor": 0.014523166658667227,
      "audioSecondsPerSecond": 68.85550675707586,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177419776,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 43.56949997600168,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": null,
      "scope": "structural",
      "cpuAgreementScoreMae": 2.1606683731079102e-7
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 42.43537498405203,
      "inferenceMs": 42.43537498405203,
      "audioSeconds": 1,
      "realTimeFactor": 0.04243537498405203,
      "audioSecondsPerSecond": 23.565244807564863,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177419776,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 42.43537498405203,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a high pitched electronic beep.",
      "scope": "top1",
      "cpuAgreementScoreMae": 2.5920599000528455e-10
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 85.74524999130517,
      "inferenceMs": 85.74524999130517,
      "audioSeconds": 12,
      "realTimeFactor": 0.007145437499275431,
      "audioSecondsPerSecond": 139.94944327781232,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 2,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177419776,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 85.74524999130517,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.2005330063402653e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 42.898125015199184,
      "inferenceMs": 42.898125015199184,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01675708008406218,
      "audioSecondsPerSecond": 59.67626788100807,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177419776,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 42.898125015199184,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 42.448332998901606,
      "inferenceMs": 42.448332998901606,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01658138007769594,
      "audioSecondsPerSecond": 60.308610942772304,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177419776,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 42.448332998901606,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 46.97404196485877,
      "inferenceMs": 46.97404196485877,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.018349235142522957,
      "audioSecondsPerSecond": 54.49818437840911,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177419776,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 46.97404196485877,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 588.4181250003166,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 273.68545898934826,
      "inferenceMs": 273.68545898934826,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.10690838241771417,
      "audioSecondsPerSecond": 9.353803484677037,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177436160,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 862.1035839896649,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 45.63729098299518,
      "inferenceMs": 45.63729098299518,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.017827066790232493,
      "audioSecondsPerSecond": 56.09447767076877,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177436160,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 45.63729098299518,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 46.21533304452896,
      "inferenceMs": 46.21533304452896,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.018052864470519125,
      "audioSecondsPerSecond": 55.39287139905306,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177436160,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 46.21533304452896,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 569.1261669853702,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 283.392500015907,
      "inferenceMs": 283.392500015907,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.11070019531871367,
      "audioSecondsPerSecond": 9.033407729055305,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177436160,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 852.5186670012772,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 44.453667011111975,
      "inferenceMs": 44.453667011111975,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.017364713676215615,
      "audioSecondsPerSecond": 57.588050033309585,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177436160,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 44.453667011111975,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 43.25520800193772,
      "inferenceMs": 43.25520800193772,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01689656562575692,
      "audioSecondsPerSecond": 59.18362477612682,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177436160,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 43.25520800193772,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 47.29254200356081,
      "inferenceMs": 47.29254200356081,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.018473649220140942,
      "audioSecondsPerSecond": 54.131156659061574,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177419776,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 47.29254200356081,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 46.0781249566935,
      "inferenceMs": 46.0781249566935,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.0179992675612084,
      "audioSecondsPerSecond": 55.55781626110035,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177419776,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 46.0781249566935,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 41.11737501807511,
      "inferenceMs": 41.11737501807511,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.01606147461643559,
      "audioSecondsPerSecond": 62.26078388697308,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177419776,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 41.11737501807511,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 43.05158299393952,
      "inferenceMs": 43.05158299393952,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.016817024607007625,
      "audioSecondsPerSecond": 59.463550976984465,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177419776,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 43.05158299393952,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "validSortedNormalizedCompleteScores": true,
      "expectedTopLabel": "The sound of a person speaking.",
      "scope": "top1",
      "cpuAgreementScoreMae": 1.928128767758608e-9
    },
    "runtime": {
      "engine": "transformers",
      "version": "5.22.1",
      "torchVersion": "2.10.0",
      "loadMs": 0,
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": {
        "mps:0/torch.float32": 532
      },
      "accelerationDetails": "Audio/text neural encoders on selected device; WAV decode/resample, mono mix and mel feature extraction on CPU",
      "durationMs": 46.28995800158009,
      "inferenceMs": 46.28995800158009,
      "audioSeconds": 2.56,
      "realTimeFactor": 0.018082014844367222,
      "audioSecondsPerSecond": 55.30357145523043,
      "activationPlacement": {
        "output": "mps:0"
      },
      "scorePlacement": "mps:0",
      "windows": 1,
      "windowSeconds": 10,
      "aggregation": "Mean window logits then softmax over supplied labels",
      "driverAllocatedMemoryBytes": 1177419776,
      "device": "mps",
      "precision": "fp32",
      "cpuThreads": 4,
      "totalMs": 46.28995800158009,
      "artifactRevision": "ada0c23a36c4e8582805bb38fec3905903f18b41",
      "fallback": false,
      "grade": "fast",
      "modelId": "clap-larger-general-native"
    }
  }
]

Something not quite right?

Tell us what happened with LAION CLAP. Loading, results, or the page itself—we’re here for all of it.

Attachments Optional

Up to 3 files · 50 MB each

Browser, OS, CPU/GPU, memory, screen, storage, and network details are included automatically. Model inputs, results, and browsing history are not collected.