{
  "schemaVersion": 1,
  "engineFamily": "whisper-file",
  "published": {
    "sanitizedAt": "2026-09-21T01:38:42.341Z",
    "sourceFileName": "whisper-cuda-win-5202a48f0-2026-09-20_15-58-15.json",
    "note": "Paths, transcripts, and local machine identifiers removed for public hosting."
  },
  "environment": {
    "collectedAt": "2026-09-20T15:58:15.903Z",
    "platform": "win32",
    "arch": "x64",
    "cpuModel": "AMD Ryzen Threadripper 3960X",
    "gpuModel": "NVIDIA GeForce RTX 3080",
    "nodeVersion": "v24.14.0",
    "gitShortSha": "5202a48f0",
    "engineFamily": "whisper-file",
    "gpuEngine": "CUDA",
    "useGpu": true,
    "threads": 4
  },
  "runConfig": {
    "engine": "whisper-file",
    "scene": "general",
    "useGpu": true,
    "gpuEngine": "CUDA",
    "threads": 4,
    "languageFromCase": true,
    "notes": "Windows + NVIDIA only. Aligns with general file-transcription preset; record whisper.cpp commit in environment."
  },
  "manifestPath": "manifest.json",
  "modelsPath": "whisper-models.json",
  "modelRuns": [
    {
      "modelId": "Base",
      "whisperModel": "Base",
      "batch": {
        "engine": "whisper-file",
        "results": [
          {
            "engine": "whisper-file",
            "model": "Base",
            "caseId": "audio-benchmark",
            "segmentCount": 10,
            "errorMetricKind": "wer",
            "wer": 0,
            "cer": null,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0,
            "duplicateTokenRate": 0,
            "audioDurationMs": 34537,
            "preprocessElapsedMs": 99,
            "transcribeElapsedMs": 2342,
            "totalElapsedMs": 2441,
            "realTimeFactor": 0.0707,
            "sceneHint": "general",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "Base",
            "caseId": "en-long-sentence-pause",
            "segmentCount": 23,
            "errorMetricKind": "wer",
            "wer": 0.0049,
            "cer": null,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0.0049,
            "duplicateTokenRate": 0,
            "audioDurationMs": 70993,
            "preprocessElapsedMs": 112,
            "transcribeElapsedMs": 2851,
            "totalElapsedMs": 2963,
            "realTimeFactor": 0.0417,
            "sceneHint": "speech",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "Base",
            "caseId": "en-oral-correction",
            "segmentCount": 12,
            "errorMetricKind": "wer",
            "wer": 0.0792,
            "cer": null,
            "insertionRate": 0.005,
            "deletionRate": 0.0347,
            "substitutionRate": 0.0396,
            "duplicateTokenRate": 0.0204,
            "audioDurationMs": 68449,
            "preprocessElapsedMs": 102,
            "transcribeElapsedMs": 2740,
            "totalElapsedMs": 2842,
            "realTimeFactor": 0.0415,
            "sceneHint": "dialogue",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "Base",
            "caseId": "en-tech-talk-podcast",
            "segmentCount": 22,
            "errorMetricKind": "wer",
            "wer": 0.012,
            "cer": null,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0.012,
            "duplicateTokenRate": 0.006,
            "audioDurationMs": 79681,
            "preprocessElapsedMs": 106,
            "transcribeElapsedMs": 2629,
            "totalElapsedMs": 2735,
            "realTimeFactor": 0.0343,
            "sceneHint": "course",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "Base",
            "caseId": "zh-tech-podcast",
            "segmentCount": 14,
            "errorMetricKind": "cer",
            "wer": null,
            "cer": 0.0258,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0.0258,
            "duplicateTokenRate": 0.5714,
            "audioDurationMs": 74665,
            "preprocessElapsedMs": 101,
            "transcribeElapsedMs": 3391,
            "totalElapsedMs": 3492,
            "realTimeFactor": 0.0468,
            "sceneHint": "course",
            "notes": ["scene=general"]
          }
        ],
        "failures": [],
        "summary": {
          "caseCount": 5,
          "successCount": 5,
          "crashCount": 0,
          "crashRate": 0,
          "averageWer": 0.024,
          "averageCer": 0.0258,
          "averageInsertionRate": 0.001,
          "averageDeletionRate": 0.0069,
          "averageSubstitutionRate": 0.0165,
          "averageDuplicateTokenRate": 0.1196,
          "averagePreprocessElapsedMs": 104,
          "averageTotalElapsedMs": 2894.6,
          "averageRealTimeFactor": 0.047,
          "byScene": {
            "general": {
              "caseCount": 1,
              "successCount": 1,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0,
              "averageCer": null,
              "averageInsertionRate": 0,
              "averageDeletionRate": 0,
              "averageSubstitutionRate": 0,
              "averageDuplicateTokenRate": 0,
              "averagePreprocessElapsedMs": 99,
              "averageTotalElapsedMs": 2441,
              "averageRealTimeFactor": 0.0707
            },
            "speech": {
              "caseCount": 1,
              "successCount": 1,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0.0049,
              "averageCer": null,
              "averageInsertionRate": 0,
              "averageDeletionRate": 0,
              "averageSubstitutionRate": 0.0049,
              "averageDuplicateTokenRate": 0,
              "averagePreprocessElapsedMs": 112,
              "averageTotalElapsedMs": 2963,
              "averageRealTimeFactor": 0.0417
            },
            "dialogue": {
              "caseCount": 1,
              "successCount": 1,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0.0792,
              "averageCer": null,
              "averageInsertionRate": 0.005,
              "averageDeletionRate": 0.0347,
              "averageSubstitutionRate": 0.0396,
              "averageDuplicateTokenRate": 0.0204,
              "averagePreprocessElapsedMs": 102,
              "averageTotalElapsedMs": 2842,
              "averageRealTimeFactor": 0.0415
            },
            "course": {
              "caseCount": 2,
              "successCount": 2,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0.012,
              "averageCer": 0.0258,
              "averageInsertionRate": 0,
              "averageDeletionRate": 0,
              "averageSubstitutionRate": 0.0189,
              "averageDuplicateTokenRate": 0.2887,
              "averagePreprocessElapsedMs": 103.5,
              "averageTotalElapsedMs": 3113.5,
              "averageRealTimeFactor": 0.0406
            }
          }
        }
      }
    },
    {
      "modelId": "Small",
      "whisperModel": "Small",
      "batch": {
        "engine": "whisper-file",
        "results": [
          {
            "engine": "whisper-file",
            "model": "Small",
            "caseId": "audio-benchmark",
            "segmentCount": 12,
            "errorMetricKind": "wer",
            "wer": 0,
            "cer": null,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0,
            "duplicateTokenRate": 0,
            "audioDurationMs": 34537,
            "preprocessElapsedMs": 96,
            "transcribeElapsedMs": 2412,
            "totalElapsedMs": 2508,
            "realTimeFactor": 0.0726,
            "sceneHint": "general",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "Small",
            "caseId": "en-long-sentence-pause",
            "segmentCount": 21,
            "errorMetricKind": "wer",
            "wer": 0,
            "cer": null,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0,
            "duplicateTokenRate": 0,
            "audioDurationMs": 70993,
            "preprocessElapsedMs": 106,
            "transcribeElapsedMs": 4149,
            "totalElapsedMs": 4255,
            "realTimeFactor": 0.0599,
            "sceneHint": "speech",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "Small",
            "caseId": "en-oral-correction",
            "segmentCount": 13,
            "errorMetricKind": "wer",
            "wer": 0.0644,
            "cer": null,
            "insertionRate": 0.005,
            "deletionRate": 0.0149,
            "substitutionRate": 0.0446,
            "duplicateTokenRate": 0.025,
            "audioDurationMs": 68449,
            "preprocessElapsedMs": 100,
            "transcribeElapsedMs": 4177,
            "totalElapsedMs": 4277,
            "realTimeFactor": 0.0625,
            "sceneHint": "dialogue",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "Small",
            "caseId": "en-tech-talk-podcast",
            "segmentCount": 12,
            "errorMetricKind": "wer",
            "wer": 0.006,
            "cer": null,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0.006,
            "duplicateTokenRate": 0.006,
            "audioDurationMs": 79681,
            "preprocessElapsedMs": 104,
            "transcribeElapsedMs": 3660,
            "totalElapsedMs": 3764,
            "realTimeFactor": 0.0472,
            "sceneHint": "course",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "Small",
            "caseId": "zh-tech-podcast",
            "segmentCount": 34,
            "errorMetricKind": "cer",
            "wer": null,
            "cer": 0.0097,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0.0097,
            "duplicateTokenRate": 0.8333,
            "audioDurationMs": 74665,
            "preprocessElapsedMs": 102,
            "transcribeElapsedMs": 4980,
            "totalElapsedMs": 5082,
            "realTimeFactor": 0.0681,
            "sceneHint": "course",
            "notes": ["scene=general"]
          }
        ],
        "failures": [],
        "summary": {
          "caseCount": 5,
          "successCount": 5,
          "crashCount": 0,
          "crashRate": 0,
          "averageWer": 0.0176,
          "averageCer": 0.0097,
          "averageInsertionRate": 0.001,
          "averageDeletionRate": 0.003,
          "averageSubstitutionRate": 0.0121,
          "averageDuplicateTokenRate": 0.1729,
          "averagePreprocessElapsedMs": 101.6,
          "averageTotalElapsedMs": 3977.2,
          "averageRealTimeFactor": 0.0621,
          "byScene": {
            "general": {
              "caseCount": 1,
              "successCount": 1,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0,
              "averageCer": null,
              "averageInsertionRate": 0,
              "averageDeletionRate": 0,
              "averageSubstitutionRate": 0,
              "averageDuplicateTokenRate": 0,
              "averagePreprocessElapsedMs": 96,
              "averageTotalElapsedMs": 2508,
              "averageRealTimeFactor": 0.0726
            },
            "speech": {
              "caseCount": 1,
              "successCount": 1,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0,
              "averageCer": null,
              "averageInsertionRate": 0,
              "averageDeletionRate": 0,
              "averageSubstitutionRate": 0,
              "averageDuplicateTokenRate": 0,
              "averagePreprocessElapsedMs": 106,
              "averageTotalElapsedMs": 4255,
              "averageRealTimeFactor": 0.0599
            },
            "dialogue": {
              "caseCount": 1,
              "successCount": 1,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0.0644,
              "averageCer": null,
              "averageInsertionRate": 0.005,
              "averageDeletionRate": 0.0149,
              "averageSubstitutionRate": 0.0446,
              "averageDuplicateTokenRate": 0.025,
              "averagePreprocessElapsedMs": 100,
              "averageTotalElapsedMs": 4277,
              "averageRealTimeFactor": 0.0625
            },
            "course": {
              "caseCount": 2,
              "successCount": 2,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0.006,
              "averageCer": 0.0097,
              "averageInsertionRate": 0,
              "averageDeletionRate": 0,
              "averageSubstitutionRate": 0.0078,
              "averageDuplicateTokenRate": 0.4197,
              "averagePreprocessElapsedMs": 103,
              "averageTotalElapsedMs": 4423,
              "averageRealTimeFactor": 0.0576
            }
          }
        }
      }
    },
    {
      "modelId": "LargeV3Turbo",
      "whisperModel": "LargeV3Turbo",
      "batch": {
        "engine": "whisper-file",
        "results": [
          {
            "engine": "whisper-file",
            "model": "LargeV3Turbo",
            "caseId": "audio-benchmark",
            "segmentCount": 6,
            "errorMetricKind": "wer",
            "wer": 0,
            "cer": null,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0,
            "duplicateTokenRate": 0,
            "audioDurationMs": 34537,
            "preprocessElapsedMs": 98,
            "transcribeElapsedMs": 2635,
            "totalElapsedMs": 2733,
            "realTimeFactor": 0.0791,
            "sceneHint": "general",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "LargeV3Turbo",
            "caseId": "en-long-sentence-pause",
            "segmentCount": 14,
            "errorMetricKind": "wer",
            "wer": 0,
            "cer": null,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0,
            "duplicateTokenRate": 0,
            "audioDurationMs": 70993,
            "preprocessElapsedMs": 98,
            "transcribeElapsedMs": 3798,
            "totalElapsedMs": 3896,
            "realTimeFactor": 0.0549,
            "sceneHint": "speech",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "LargeV3Turbo",
            "caseId": "en-oral-correction",
            "segmentCount": 32,
            "errorMetricKind": "wer",
            "wer": 0.0743,
            "cer": null,
            "insertionRate": 0.005,
            "deletionRate": 0.0198,
            "substitutionRate": 0.0495,
            "duplicateTokenRate": 0.0201,
            "audioDurationMs": 68449,
            "preprocessElapsedMs": 97,
            "transcribeElapsedMs": 4024,
            "totalElapsedMs": 4121,
            "realTimeFactor": 0.0602,
            "sceneHint": "dialogue",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "LargeV3Turbo",
            "caseId": "en-tech-talk-podcast",
            "segmentCount": 11,
            "errorMetricKind": "wer",
            "wer": 0.006,
            "cer": null,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0.006,
            "duplicateTokenRate": 0.006,
            "audioDurationMs": 79681,
            "preprocessElapsedMs": 100,
            "transcribeElapsedMs": 3544,
            "totalElapsedMs": 3644,
            "realTimeFactor": 0.0457,
            "sceneHint": "course",
            "notes": ["scene=general"]
          },
          {
            "engine": "whisper-file",
            "model": "LargeV3Turbo",
            "caseId": "zh-tech-podcast",
            "segmentCount": 13,
            "errorMetricKind": "cer",
            "wer": null,
            "cer": 0.0097,
            "insertionRate": 0,
            "deletionRate": 0,
            "substitutionRate": 0.0097,
            "duplicateTokenRate": 0.5714,
            "audioDurationMs": 74665,
            "preprocessElapsedMs": 106,
            "transcribeElapsedMs": 4300,
            "totalElapsedMs": 4406,
            "realTimeFactor": 0.059,
            "sceneHint": "course",
            "notes": ["scene=general"]
          }
        ],
        "failures": [],
        "summary": {
          "caseCount": 5,
          "successCount": 5,
          "crashCount": 0,
          "crashRate": 0,
          "averageWer": 0.0201,
          "averageCer": 0.0097,
          "averageInsertionRate": 0.001,
          "averageDeletionRate": 0.004,
          "averageSubstitutionRate": 0.013,
          "averageDuplicateTokenRate": 0.1195,
          "averagePreprocessElapsedMs": 99.8,
          "averageTotalElapsedMs": 3760,
          "averageRealTimeFactor": 0.0598,
          "byScene": {
            "general": {
              "caseCount": 1,
              "successCount": 1,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0,
              "averageCer": null,
              "averageInsertionRate": 0,
              "averageDeletionRate": 0,
              "averageSubstitutionRate": 0,
              "averageDuplicateTokenRate": 0,
              "averagePreprocessElapsedMs": 98,
              "averageTotalElapsedMs": 2733,
              "averageRealTimeFactor": 0.0791
            },
            "speech": {
              "caseCount": 1,
              "successCount": 1,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0,
              "averageCer": null,
              "averageInsertionRate": 0,
              "averageDeletionRate": 0,
              "averageSubstitutionRate": 0,
              "averageDuplicateTokenRate": 0,
              "averagePreprocessElapsedMs": 98,
              "averageTotalElapsedMs": 3896,
              "averageRealTimeFactor": 0.0549
            },
            "dialogue": {
              "caseCount": 1,
              "successCount": 1,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0.0743,
              "averageCer": null,
              "averageInsertionRate": 0.005,
              "averageDeletionRate": 0.0198,
              "averageSubstitutionRate": 0.0495,
              "averageDuplicateTokenRate": 0.0201,
              "averagePreprocessElapsedMs": 97,
              "averageTotalElapsedMs": 4121,
              "averageRealTimeFactor": 0.0602
            },
            "course": {
              "caseCount": 2,
              "successCount": 2,
              "crashCount": 0,
              "crashRate": 0,
              "averageWer": 0.006,
              "averageCer": 0.0097,
              "averageInsertionRate": 0,
              "averageDeletionRate": 0,
              "averageSubstitutionRate": 0.0078,
              "averageDuplicateTokenRate": 0.2887,
              "averagePreprocessElapsedMs": 103,
              "averageTotalElapsedMs": 4025,
              "averageRealTimeFactor": 0.0523
            }
          }
        }
      }
    },
    {
      "modelId": "LargeV3",
      "whisperModel": "LargeV3",
      "batch": {
        "engine": "whisper-file",
        "results": [],
        "failures": [],
        "summary": {
          "caseCount": 0,
          "successCount": 0,
          "crashCount": 0,
          "crashRate": 0,
          "averageWer": null,
          "averageCer": null,
          "averageInsertionRate": null,
          "averageDeletionRate": null,
          "averageSubstitutionRate": null,
          "averageDuplicateTokenRate": 0,
          "averagePreprocessElapsedMs": 0,
          "averageTotalElapsedMs": 0,
          "averageRealTimeFactor": null,
          "byScene": {}
        }
      }
    },
    {
      "modelId": "BelleLargeV3TurboChinese",
      "whisperModel": "BelleLargeV3TurboChinese",
      "batch": {
        "engine": "whisper-file",
        "results": [],
        "failures": [],
        "summary": {
          "caseCount": 0,
          "successCount": 0,
          "crashCount": 0,
          "crashRate": 0,
          "averageWer": null,
          "averageCer": null,
          "averageInsertionRate": null,
          "averageDeletionRate": null,
          "averageSubstitutionRate": null,
          "averageDuplicateTokenRate": 0,
          "averagePreprocessElapsedMs": 0,
          "averageTotalElapsedMs": 0,
          "averageRealTimeFactor": null,
          "byScene": {}
        }
      }
    }
  ],
  "skipped": [
    {
      "modelId": "LargeV3",
      "caseId": "audio-benchmark",
      "reason": "whisper_model_missing"
    },
    {
      "modelId": "LargeV3",
      "caseId": "en-long-sentence-pause",
      "reason": "whisper_model_missing"
    },
    {
      "modelId": "LargeV3",
      "caseId": "en-oral-correction",
      "reason": "whisper_model_missing"
    },
    {
      "modelId": "LargeV3",
      "caseId": "en-tech-talk-podcast",
      "reason": "whisper_model_missing"
    },
    {
      "modelId": "LargeV3",
      "caseId": "zh-tech-podcast",
      "reason": "whisper_model_missing"
    },
    {
      "modelId": "BelleLargeV3TurboChinese",
      "caseId": "audio-benchmark",
      "reason": "language_not_supported"
    },
    {
      "modelId": "BelleLargeV3TurboChinese",
      "caseId": "en-long-sentence-pause",
      "reason": "language_not_supported"
    },
    {
      "modelId": "BelleLargeV3TurboChinese",
      "caseId": "en-oral-correction",
      "reason": "language_not_supported"
    },
    {
      "modelId": "BelleLargeV3TurboChinese",
      "caseId": "en-tech-talk-podcast",
      "reason": "language_not_supported"
    },
    {
      "modelId": "BelleLargeV3TurboChinese",
      "caseId": "zh-tech-podcast",
      "reason": "whisper_model_missing"
    }
  ]
}
