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.

Silero VAD

Detect speech regions in 16 kHz mono audio and return start/end timestamps locally. CPU or explicit Mac GPU execution, with CPU fallback.

Desktop inferencePrivate · Runs on your computer
About this model

Detect speech regions in 16 kHz mono audio and return start/end timestamps locally. CPU or explicit Mac GPU execution, with CPU fallback.

Created by
Silero
License
MIT
Model released
Unknown
Source repository created
Nov 23, 2020
Model weights · FP32
2.2 MB

Best for

  • Finding speech segments
  • Preparing audio for transcription
  • Detecting silence boundaries

Keep in mind

  • Not transcription, speaker identification or word alignment.
  • Music, whispers, noise and overlapping voices can cause missed speech or false detections.
  • Not a safety-critical listening or surveillance decision system.
  • Input must be normalized mono 16 kHz PCM. App decodes and resamples recordings; available memory limits capacity.
  • Windows CUDA is implemented but unverified; Intel Mac and other OS versions are unverified.
  • The app downloads its own pinned CPython and isolated packages on request; installed Python, user packages and Python/pip configuration are not used. Automatic model-package provisioning is verified on Apple Silicon Mac; other platform stacks remain unverified.

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 · cpu (verified platform only)

Sources & technical details
Original model
https://github.com/snakers4/silero-vad
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
Oct 22, 2026
Native model files
https://github.com/snakers4/silero-vad
Files updated
Oct 22, 2026
Artifact revision
066409f3d9db45a30d21dfcbf9147826ffdd7052

Desktop app

Sample result

Silero VAD — Speech Detection

Run Silero VAD 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.

Silero VAD — Speech Detection

63 configurations tested
Passed63
Failed0
TestedNov 1, 2026
Runtimepytorch 2.6.0

Quality criteria

{ "version": 1, "source": "model-samples/desktop/silero-speech.wav", "provenance": "16kHz derivative of retained Chatterbox sample, project-authored text and generated built-in voice; watermark retained. Not human-annotated natural speech.", "fixtures": [ { "id": "speech", "transform": "identity", "minimumSpeechCoverage": 0.5 }, { "id": "padded-two-utterances", "transform": "padded-repeat", "minimumSpeechCoverage": 0.5, "maximumSilenceCoverage": 0.15, "minimumSegments": 2 }, { "id": "silence", "transform": "zeros", "maximumSegments": 0 }, { "id": "low-noise", "transform": "seeded-noise", "maximumSegments": 0 } ], "gpuProbabilityMaxAbsoluteError": 0.02, "note": "Broad known placement/negative gates are frozen before results. This is limited compatibility quality, not a general VAD accuracy evaluation." }

PrecisionDeviceResultPlatformInference / loadTask metric
FP32mpsAcceleratedPassedVery fastmacos · aarch64142 ms / 575 ms0.047 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64143 ms / 545 ms0.048 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64145 ms / 572 ms0.048 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64427 ms / 545 ms0.048 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64142 ms / 545 ms0.049 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64431 ms / 575 ms0.049 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64147 ms / 545 ms0.049 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64143 ms / 545 ms0.049 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64437 ms / 572 ms0.049 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64145 ms / 572 ms0.050 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64149 ms / 575 ms0.050 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64146 ms / 572 ms0.050 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64146 ms / 575 ms0.050 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64146 ms / 575 ms0.050 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64146 ms / 572 ms0.050 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64146 ms / 545 ms0.050 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64150 ms / 572 ms0.050 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64146 ms / 545 ms0.050 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64148 ms / 575 ms0.051 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64149 ms / 572 ms0.051 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64153 ms / 545 ms0.052 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64153 ms / 575 ms0.052 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64156 ms / 572 ms0.053 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64168 ms / 575 ms0.058 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64236 ms / 545 ms0.081 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64242 ms / 572 ms0.083 RTF
FP32mpsAcceleratedPassedVery fastmacos · aarch64260 ms / 575 ms0.089 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 550 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 542 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6428 ms / 550 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 542 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 542 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 550 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 550 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6428 ms / 542 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6428 ms / 536 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 536 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 536 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 542 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 542 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 550 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 550 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 542 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 550 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch649 ms / 542 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 536 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 550 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 601 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 536 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 601 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 536 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 536 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6429 ms / 601 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 601 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 601 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 601 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6410 ms / 601 ms0.003 RTF
FP32cpuPassedVery fastmacos · aarch6411 ms / 536 ms0.004 RTF
FP32cpuPassedVery fastmacos · aarch6413 ms / 601 ms0.004 RTF
FP32cpuPassedVery fastmacos · aarch6444 ms / 542 ms0.015 RTF
FP32cpuPassedVery fastmacos · aarch6444 ms / 550 ms0.015 RTF
FP32cpuPassedVery fastmacos · aarch6445 ms / 536 ms0.015 RTF
FP32cpuPassedVery fastmacos · aarch6447 ms / 601 ms0.016 RTF
Test setup & sample details
Quality gate
{ "version": 1, "source": "model-samples/desktop/silero-speech.wav", "provenance": "16kHz derivative of retained Chatterbox sample, project-authored text and generated built-in voice; watermark retained. Not human-annotated natural speech.", "fixtures": [ { "id": "speech", "transform": "identity", "minimumSpeechCoverage": 0.5 }, { "id": "padded-two-utterances", "transform": "padded-repeat", "minimumSpeechCoverage": 0.5, "maximumSilenceCoverage": 0.15, "minimumSegments": 2 }, { "id": "silence", "transform": "zeros", "maximumSegments": 0 }, { "id": "low-noise", "transform": "seeded-noise", "maximumSegments": 0 } ], "gpuProbabilityMaxAbsoluteError": 0.02, "note": "Broad known placement/negative gates are frozen before results. This is limited compatibility quality, not a general VAD accuracy evaluation." }
Tested
2026-11-01T03:03:16Z
Environment
{ "os": "macOS 27.2", "arch": "aarch64", "hardware": "Apple M3 Max", "scope": "Generated utterance and deterministic derivatives/negatives; limited compatibility, not natural-speech accuracy certification" }
Runtime
pytorch 2.6.0
Artifact revision
066409f3d9db45a30d21dfcbf9147826ffdd7052
Fixture
{
  "file": "model-samples/desktop/silero-speech.wav",
  "url": "https://assets.localai.tools/samples/v1/files/c6b38bb2aeabdd4334ddab2c402f52c072f1f7286a28148ed336d147546fd7f3/silero-speech.wav",
  "size": 93518,
  "sha1": "2ba325d33b310957760321e7378dc328315278e4",
  "sha256": "c6b38bb2aeabdd4334ddab2c402f52c072f1f7286a28148ed336d147546fd7f3"
}
Rerun when
Artifact/runtime/preprocessing/config/host changes; Fixture checksum or frozen quality gate changes
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 600.7263329811394,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 47.33916599070653,
      "inferenceMs": 47.32245800551027,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.01620632123476379,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 648.065498971846,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 600.7263329811394,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.98208299279213,
      "inferenceMs": 9.96858297730796,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0034138982798999863,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 610.7084159739316,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 600.7263329811394,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 12.973083008546382,
      "inferenceMs": 12.956624996149912,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.00443720034114723,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 613.6994159896858,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 600.7263329811394,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 10.17008299822919,
      "inferenceMs": 10.153875016840175,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0034773544578219776,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 610.8964159793686,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 600.7263329811394,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.795375022804365,
      "inferenceMs": 9.778750012628734,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0033488869906262786,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 610.5217080039438,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 600.7263329811394,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 10.224291996564716,
      "inferenceMs": 10.203999991063029,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.003494520544884599,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 610.9506249777041,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8684931506849315,
      "silenceCoverage": 0,
      "segments": 4,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 600.7263329811394,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 29.400500003248453,
      "inferenceMs": 29.370625008596107,
      "audioSeconds": 8.84,
      "realTimeFactor": 0.0033224688923751255,
      "frames": 277,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 630.1268329843879,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 600.7263329811394,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.841582999797538,
      "inferenceMs": 9.834500000579283,
      "audioSeconds": 3,
      "realTimeFactor": 0.003278166666859761,
      "frames": 94,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 610.567915980937,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 600.7263329811394,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.850417001871392,
      "inferenceMs": 9.843667008681223,
      "audioSeconds": 3,
      "realTimeFactor": 0.0032812223362270743,
      "frames": 94,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 610.5767499830108,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 542.2940410207957,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 43.573542003287,
      "inferenceMs": 43.545541993808,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.014912856847194522,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 585.8675830240827,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 542.2940410207957,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.510207979474217,
      "inferenceMs": 9.48991600307636,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.00324997123393026,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 551.8042490002699,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 542.2940410207957,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.241417021257803,
      "inferenceMs": 9.225042012985796,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0031592609633513,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 551.5354580420535,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 542.2940410207957,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.17033400037326,
      "inferenceMs": 9.155583975370973,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0031354739641681418,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 551.464375021169,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 542.2940410207957,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.376166999572888,
      "inferenceMs": 9.358834009617567,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.003205080140279989,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 551.6702080203686,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 542.2940410207957,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.340667020296678,
      "inferenceMs": 9.3205840094015,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0031919808251375,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 551.6347080410924,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8684931506849315,
      "silenceCoverage": 0,
      "segments": 4,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 542.2940410207957,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 28.15033300430514,
      "inferenceMs": 28.12187501695007,
      "audioSeconds": 8.84,
      "realTimeFactor": 0.0031812075811029493,
      "frames": 277,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 570.4443740251008,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 542.2940410207957,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.478249994572252,
      "inferenceMs": 9.47124999947846,
      "audioSeconds": 3,
      "realTimeFactor": 0.0031570833331594863,
      "frames": 94,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 551.772291015368,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 542.2940410207957,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.577082993928343,
      "inferenceMs": 9.571665985276923,
      "audioSeconds": 3,
      "realTimeFactor": 0.0031905553284256407,
      "frames": 94,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 551.871124014724,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 536.3427090051118,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 45.18462499254383,
      "inferenceMs": 45.14829098479822,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.015461743487944596,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 581.5273339976557,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 536.3427090051118,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.594875009497628,
      "inferenceMs": 9.576333017321303,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.003279566101822364,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 545.9375840146095,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 536.3427090051118,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.68541597831063,
      "inferenceMs": 9.668874990893528,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0033112585585251806,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 546.0281249834225,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 536.3427090051118,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.530042007099837,
      "inferenceMs": 9.513583005173132,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.003258076371634634,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 545.8727510122117,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 536.3427090051118,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.315374976722524,
      "inferenceMs": 9.296125004766509,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.003183604453687161,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 545.6580839818344,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 536.3427090051118,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 10.942915978375822,
      "inferenceMs": 10.919957974692807,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0037397116351687693,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 547.2856249834877,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8684931506849315,
      "silenceCoverage": 0,
      "segments": 4,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 536.3427090051118,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 28.16866699140519,
      "inferenceMs": 28.13762499135919,
      "audioSeconds": 8.84,
      "realTimeFactor": 0.003182989252416198,
      "frames": 277,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 564.511375996517,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 536.3427090051118,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.576124983141199,
      "inferenceMs": 9.569166984874755,
      "audioSeconds": 3,
      "realTimeFactor": 0.003189722328291585,
      "frames": 94,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 545.918833988253,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 536.3427090051118,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.856165997916833,
      "inferenceMs": 9.850041009485722,
      "audioSeconds": 3,
      "realTimeFactor": 0.003283347003161907,
      "frames": 94,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 546.1988750030287,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 550.4136249946896,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 43.79062500083819,
      "inferenceMs": 43.77070799819194,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.014989968492531485,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 2,
      "totalMs": 594.2042499955278,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 550.4136249946896,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.246667003026232,
      "inferenceMs": 9.228667011484504,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0031605024011933233,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 2,
      "totalMs": 559.6602919977158,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 550.4136249946896,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.342208999441937,
      "inferenceMs": 9.322917001554742,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.003192779795052994,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 2,
      "totalMs": 559.7558339941315,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 550.4136249946896,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.301458980189636,
      "inferenceMs": 9.28204198135063,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0031787815004625446,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 2,
      "totalMs": 559.7150839748792,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 550.4136249946896,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.47195800836198,
      "inferenceMs": 9.450457990169525,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0032364582158114813,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 2,
      "totalMs": 559.8855830030516,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 550.4136249946896,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.115749999182299,
      "inferenceMs": 9.099541988689452,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.00311628150297584,
      "frames": 92,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 2,
      "totalMs": 559.5293749938719,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8684931506849315,
      "silenceCoverage": 0,
      "segments": 4,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 550.4136249946896,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 27.758957992773503,
      "inferenceMs": 27.734041010262445,
      "audioSeconds": 8.84,
      "realTimeFactor": 0.0031373349559120412,
      "frames": 277,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 2,
      "totalMs": 578.1725829874631,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 550.4136249946896,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.825291985180229,
      "inferenceMs": 9.819208993576467,
      "audioSeconds": 3,
      "realTimeFactor": 0.003273069664525489,
      "frames": 94,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 2,
      "totalMs": 560.2389169798698,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 550.4136249946896,
      "device": "cpu",
      "provider": "PyTorch CPU",
      "acceleration": "cpu",
      "placement": "cpu",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 9.584915998857468,
      "inferenceMs": 9.578375000273809,
      "audioSeconds": 3,
      "realTimeFactor": 0.003192791666757936,
      "frames": 94,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 2,
      "totalMs": 559.9985409935471,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 574.5475000003353,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 260.14883298194036,
      "inferenceMs": 260.11816697428003,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.08908156403228767,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 834.6963329822756,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 574.5475000003353,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 153.1799160002265,
      "inferenceMs": 153.15920798457228,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.05245178355636037,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 727.7274160005618,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 574.5475000003353,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 168.27416699379683,
      "inferenceMs": 168.25337501359172,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.05762101884027114,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 742.8216669941321,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 574.5475000003353,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 148.22470798389986,
      "inferenceMs": 148.20362499449402,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.0507546660940048,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 722.7722079842351,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 574.5475000003353,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 145.889375009574,
      "inferenceMs": 145.86854199296795,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.04995498013457807,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 720.4368750099093,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 574.5475000003353,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 145.56091601843946,
      "inferenceMs": 145.53716601221822,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.04984149520966377,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 720.1084160187747,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8684931506849315,
      "silenceCoverage": 0,
      "segments": 4,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 574.5475000003353,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 431.09875000664033,
      "inferenceMs": 431.06333300238475,
      "audioSeconds": 8.84,
      "realTimeFactor": 0.04876282047538289,
      "frames": 277,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 1005.6462500069756,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 574.5475000003353,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 142.38270901842043,
      "inferenceMs": 142.37445901380852,
      "audioSeconds": 3,
      "realTimeFactor": 0.04745815300460284,
      "frames": 94,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 716.9302090187557,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 574.5475000003353,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 148.84220901876688,
      "inferenceMs": 148.83395901415497,
      "audioSeconds": 3,
      "realTimeFactor": 0.04961131967138499,
      "frames": 94,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 723.3897090191022,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 544.776249997085,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 236.05791700538248,
      "inferenceMs": 236.03500000899658,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.08083390411267007,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 780.8341670024674,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 544.776249997085,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 142.26487500127405,
      "inferenceMs": 142.24533300148323,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.048714155137494254,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 687.041124998359,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 544.776249997085,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 146.29741699900478,
      "inferenceMs": 146.27816699794494,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.050095262670529085,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 691.0736669960897,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 544.776249997085,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 153.11445901170373,
      "inferenceMs": 153.09450001223013,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.052429623291859635,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 697.8907090087887,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 544.776249997085,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 145.9699160186574,
      "inferenceMs": 145.94933300395496,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.04998264828902567,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 690.7461660157423,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 544.776249997085,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 143.2652910007164,
      "inferenceMs": 143.2452910230495,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.04905660651474298,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 688.0415409978013,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8684931506849315,
      "silenceCoverage": 0,
      "segments": 4,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 544.776249997085,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 426.8836670089513,
      "inferenceMs": 426.8544169899542,
      "audioSeconds": 8.84,
      "realTimeFactor": 0.04828669875451971,
      "frames": 277,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 971.6599170060363,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 544.776249997085,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 143.13112499075942,
      "inferenceMs": 143.1229170120787,
      "audioSeconds": 3,
      "realTimeFactor": 0.04770763900402623,
      "frames": 94,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 687.9073749878444,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 544.776249997085,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 146.79287499166094,
      "inferenceMs": 146.78533299593255,
      "audioSeconds": 3,
      "realTimeFactor": 0.048928444331977516,
      "frames": 94,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 691.5691249887459,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 572.0027920033317,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 241.9031249883119,
      "inferenceMs": 241.87979201087728,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.08283554520920455,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 813.9059169916436,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 572.0027920033317,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 148.76666700001806,
      "inferenceMs": 148.744792008074,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.05093999726303904,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 720.7694590033498,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 572.0027920033317,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 155.56283298064955,
      "inferenceMs": 155.54149998934008,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.05326763698265071,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 727.5656249839813,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 572.0027920033317,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 145.95895798993297,
      "inferenceMs": 145.92845799052157,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.049975499311822456,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 717.9617499932647,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 572.0027920033317,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 144.77616699878126,
      "inferenceMs": 144.75645899074152,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.049574129791349834,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28147712,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 716.778959002113,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8883561643835617,
      "silenceCoverage": null,
      "segments": 2,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 572.0027920033317,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 145.52408302552067,
      "inferenceMs": 145.50283300923184,
      "audioSeconds": 2.92,
      "realTimeFactor": 0.04982973733192871,
      "frames": 92,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 717.5268750288524,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": 0.8684931506849315,
      "silenceCoverage": 0,
      "segments": 4,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 572.0027920033317,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 436.7115840141196,
      "inferenceMs": 436.68170901946723,
      "audioSeconds": 8.84,
      "realTimeFactor": 0.0493983833732429,
      "frames": 277,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 1008.7143760174513,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 572.0027920033317,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 144.7274579841178,
      "inferenceMs": 144.71937497728504,
      "audioSeconds": 3,
      "realTimeFactor": 0.04823979165909501,
      "frames": 94,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 716.7302499874495,
      "grade": "excellent",
      "fallback": false
    }
  },
  {
    "precision": "fp32",
    "device": "mps",
    "status": "pass",
    "quality": {
      "passed": true,
      "speechCoverage": null,
      "silenceCoverage": 0,
      "segments": 0,
      "method": "Known synthetic placement plus negative segments; frozen thresholds in quality-gates.json"
    },
    "runtime": {
      "engine": "pytorch",
      "version": "2.6.0",
      "loadMs": 572.0027920033317,
      "device": "mps",
      "provider": "Metal/MPS",
      "acceleration": "partial",
      "placement": "mps:0",
      "parameterDtype": "torch.float32",
      "accelerationDetails": "Model frames and recurrent state on selected device; PCM framing, probability extraction and segmentation on CPU",
      "durationMs": 150.18570801476017,
      "inferenceMs": 150.16449999529868,
      "audioSeconds": 3,
      "realTimeFactor": 0.050054833331766226,
      "frames": 94,
      "driverAllocatedMemoryBytes": 28131328,
      "artifactRevision": "066409f3d9db45a30d21dfcbf9147826ffdd7052",
      "precision": "fp32",
      "cpuThreads": 1,
      "totalMs": 722.1885000180919,
      "grade": "excellent",
      "fallback": false
    }
  }
]

Something not quite right?

Tell us what happened with Silero VAD. 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.