{ "schemaVersion": 1, "engineFamily": "ml-engine-asr", "published": { "sanitizedAt": "2026-09-21T03:13:30.754Z", "sourceFileName": "ml-engine-win-fe2d7bcd3-2026-09-21_03-01-22.json", "note": "Paths, transcripts, and local machine identifiers removed for public hosting." }, "environment": { "collectedAt": "2026-09-21T03:01:22.227Z", "platform": "win32", "arch": "x64", "cpuModel": "AMD Ryzen Threadripper 3960X", "gpuModel": "NVIDIA GeForce RTX 3080 (not used for ml-engine CPU run)", "nodeVersion": "v24.14.0", "gitShortSha": "fe2d7bcd3", "engineFamily": "ml-engine-asr", "provider": "cpu", "numThreads": 8 }, "runConfig": { "provider": "cpu", "numThreads": 8, "languageFromCase": true, "notes": "Default matches App file-transcription ml-engine path (CPU). Override numThreads only via CLI if needed." }, "manifestPath": "manifest.json", "modelsPath": "models.json", "modelRuns": [ { "modelId": "en-zipformer-onnx", "modelArch": "transducer", "batch": { "engine": "ml-engine-asr", "results": [ { "engine": "ml-engine-asr", "model": "transducer", "caseId": "audio-benchmark", "segmentCount": 15, "errorMetricKind": "wer", "wer": 0.043, "cer": null, "insertionRate": 0.0108, "deletionRate": 0, "substitutionRate": 0.0323, "duplicateTokenRate": 0, "audioDurationMs": 34536, "preprocessElapsedMs": 0, "transcribeElapsedMs": 4488, "totalElapsedMs": 4488, "realTimeFactor": 0.1299, "sceneHint": "general" }, { "engine": "ml-engine-asr", "model": "transducer", "caseId": "en-long-sentence-pause", "segmentCount": 33, "errorMetricKind": "wer", "wer": 0.1133, "cer": null, "insertionRate": 0.0099, "deletionRate": 0, "substitutionRate": 0.1034, "duplicateTokenRate": 0, "audioDurationMs": 70992, "preprocessElapsedMs": 0, "transcribeElapsedMs": 5648, "totalElapsedMs": 5648, "realTimeFactor": 0.0796, "sceneHint": "speech" }, { "engine": "ml-engine-asr", "model": "transducer", "caseId": "en-oral-correction", "segmentCount": 35, "errorMetricKind": "wer", "wer": 0.0941, "cer": null, "insertionRate": 0.0248, "deletionRate": 0.0198, "substitutionRate": 0.0495, "duplicateTokenRate": 0.0148, "audioDurationMs": 68448, "preprocessElapsedMs": 0, "transcribeElapsedMs": 5657, "totalElapsedMs": 5657, "realTimeFactor": 0.0826, "sceneHint": "dialogue" }, { "engine": "ml-engine-asr", "model": "transducer", "caseId": "en-tech-talk-podcast", "segmentCount": 26, "errorMetricKind": "wer", "wer": 0.1018, "cer": null, "insertionRate": 0.024, "deletionRate": 0.012, "substitutionRate": 0.0659, "duplicateTokenRate": 0.0059, "audioDurationMs": 79680, "preprocessElapsedMs": 0, "transcribeElapsedMs": 5578, "totalElapsedMs": 5578, "realTimeFactor": 0.07, "sceneHint": "course" } ], "failures": [], "summary": { "caseCount": 4, "successCount": 4, "crashCount": 0, "crashRate": 0, "averageWer": 0.0881, "averageCer": null, "averageInsertionRate": 0.0174, "averageDeletionRate": 0.008, "averageSubstitutionRate": 0.0628, "averageDuplicateTokenRate": 0.0052, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 5342.75, "averageRealTimeFactor": 0.0905, "byScene": { "general": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.043, "averageCer": null, "averageInsertionRate": 0.0108, "averageDeletionRate": 0, "averageSubstitutionRate": 0.0323, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 4488, "averageRealTimeFactor": 0.1299 }, "speech": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.1133, "averageCer": null, "averageInsertionRate": 0.0099, "averageDeletionRate": 0, "averageSubstitutionRate": 0.1034, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 5648, "averageRealTimeFactor": 0.0796 }, "dialogue": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0941, "averageCer": null, "averageInsertionRate": 0.0248, "averageDeletionRate": 0.0198, "averageSubstitutionRate": 0.0495, "averageDuplicateTokenRate": 0.0148, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 5657, "averageRealTimeFactor": 0.0826 }, "course": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.1018, "averageCer": null, "averageInsertionRate": 0.024, "averageDeletionRate": 0.012, "averageSubstitutionRate": 0.0659, "averageDuplicateTokenRate": 0.0059, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 5578, "averageRealTimeFactor": 0.07 } } } } }, { "modelId": "zh-en-zipformer-onnx", "modelArch": "transducer", "batch": { "engine": "ml-engine-asr", "results": [ { "engine": "ml-engine-asr", "model": "transducer", "caseId": "audio-benchmark", "segmentCount": 15, "errorMetricKind": "wer", "wer": 0.0645, "cer": null, "insertionRate": 0.0215, "deletionRate": 0, "substitutionRate": 0.043, "duplicateTokenRate": 0, "audioDurationMs": 34536, "preprocessElapsedMs": 0, "transcribeElapsedMs": 3368, "totalElapsedMs": 3368, "realTimeFactor": 0.0975, "sceneHint": "general" }, { "engine": "ml-engine-asr", "model": "transducer", "caseId": "en-long-sentence-pause", "segmentCount": 33, "errorMetricKind": "wer", "wer": 0.0739, "cer": null, "insertionRate": 0.0099, "deletionRate": 0.0049, "substitutionRate": 0.0591, "duplicateTokenRate": 0, "audioDurationMs": 70992, "preprocessElapsedMs": 0, "transcribeElapsedMs": 4370, "totalElapsedMs": 4370, "realTimeFactor": 0.0616, "sceneHint": "speech" }, { "engine": "ml-engine-asr", "model": "transducer", "caseId": "en-oral-correction", "segmentCount": 35, "errorMetricKind": "wer", "wer": 0.1188, "cer": null, "insertionRate": 0.0248, "deletionRate": 0.0248, "substitutionRate": 0.0693, "duplicateTokenRate": 0.0099, "audioDurationMs": 68448, "preprocessElapsedMs": 0, "transcribeElapsedMs": 4327, "totalElapsedMs": 4327, "realTimeFactor": 0.0632, "sceneHint": "dialogue" }, { "engine": "ml-engine-asr", "model": "transducer", "caseId": "en-tech-talk-podcast", "segmentCount": 26, "errorMetricKind": "wer", "wer": 0.1377, "cer": null, "insertionRate": 0.024, "deletionRate": 0.018, "substitutionRate": 0.0958, "duplicateTokenRate": 0.006, "audioDurationMs": 79680, "preprocessElapsedMs": 0, "transcribeElapsedMs": 4190, "totalElapsedMs": 4190, "realTimeFactor": 0.0526, "sceneHint": "course" }, { "engine": "ml-engine-asr", "model": "transducer", "caseId": "zh-tech-podcast", "segmentCount": 29, "errorMetricKind": "cer", "wer": null, "cer": 0.0323, "insertionRate": 0, "deletionRate": 0.0032, "substitutionRate": 0.029, "duplicateTokenRate": 0, "audioDurationMs": 74664, "preprocessElapsedMs": 0, "transcribeElapsedMs": 4294, "totalElapsedMs": 4294, "realTimeFactor": 0.0575, "sceneHint": "course" } ], "failures": [], "summary": { "caseCount": 5, "successCount": 5, "crashCount": 0, "crashRate": 0, "averageWer": 0.0987, "averageCer": 0.0323, "averageInsertionRate": 0.016, "averageDeletionRate": 0.0102, "averageSubstitutionRate": 0.0592, "averageDuplicateTokenRate": 0.0032, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 4109.8, "averageRealTimeFactor": 0.0665, "byScene": { "general": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0645, "averageCer": null, "averageInsertionRate": 0.0215, "averageDeletionRate": 0, "averageSubstitutionRate": 0.043, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 3368, "averageRealTimeFactor": 0.0975 }, "speech": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0739, "averageCer": null, "averageInsertionRate": 0.0099, "averageDeletionRate": 0.0049, "averageSubstitutionRate": 0.0591, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 4370, "averageRealTimeFactor": 0.0616 }, "dialogue": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.1188, "averageCer": null, "averageInsertionRate": 0.0248, "averageDeletionRate": 0.0248, "averageSubstitutionRate": 0.0693, "averageDuplicateTokenRate": 0.0099, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 4327, "averageRealTimeFactor": 0.0632 }, "course": { "caseCount": 2, "successCount": 2, "crashCount": 0, "crashRate": 0, "averageWer": 0.1377, "averageCer": 0.0323, "averageInsertionRate": 0.012, "averageDeletionRate": 0.0106, "averageSubstitutionRate": 0.0624, "averageDuplicateTokenRate": 0.003, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 4242, "averageRealTimeFactor": 0.0551 } } } } }, { "modelId": "qwen3-asr", "modelArch": "qwen3-asr", "batch": { "engine": "ml-engine-asr", "results": [ { "engine": "ml-engine-asr", "model": "qwen3-asr", "caseId": "audio-benchmark", "segmentCount": 15, "errorMetricKind": "wer", "wer": 0, "cer": null, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0, "duplicateTokenRate": 0, "audioDurationMs": 34536, "preprocessElapsedMs": 0, "transcribeElapsedMs": 12983, "totalElapsedMs": 12983, "realTimeFactor": 0.3759, "sceneHint": "general" }, { "engine": "ml-engine-asr", "model": "qwen3-asr", "caseId": "en-long-sentence-pause", "segmentCount": 33, "errorMetricKind": "wer", "wer": 0, "cer": null, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0, "duplicateTokenRate": 0, "audioDurationMs": 70992, "preprocessElapsedMs": 0, "transcribeElapsedMs": 21946, "totalElapsedMs": 21946, "realTimeFactor": 0.3091, "sceneHint": "speech" }, { "engine": "ml-engine-asr", "model": "qwen3-asr", "caseId": "en-oral-correction", "segmentCount": 35, "errorMetricKind": "wer", "wer": 0.0396, "cer": null, "insertionRate": 0.005, "deletionRate": 0.0149, "substitutionRate": 0.0198, "duplicateTokenRate": 0.025, "audioDurationMs": 68448, "preprocessElapsedMs": 0, "transcribeElapsedMs": 21858, "totalElapsedMs": 21858, "realTimeFactor": 0.3193, "sceneHint": "dialogue" }, { "engine": "ml-engine-asr", "model": "qwen3-asr", "caseId": "en-tech-talk-podcast", "segmentCount": 26, "errorMetricKind": "wer", "wer": 0.024, "cer": null, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0.024, "duplicateTokenRate": 0.006, "audioDurationMs": 79680, "preprocessElapsedMs": 0, "transcribeElapsedMs": 18921, "totalElapsedMs": 18921, "realTimeFactor": 0.2375, "sceneHint": "course" }, { "engine": "ml-engine-asr", "model": "qwen3-asr", "caseId": "zh-tech-podcast", "segmentCount": 29, "errorMetricKind": "cer", "wer": null, "cer": 0.0065, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0.0065, "duplicateTokenRate": 0.8333, "audioDurationMs": 74664, "preprocessElapsedMs": 0, "transcribeElapsedMs": 19884, "totalElapsedMs": 19884, "realTimeFactor": 0.2663, "sceneHint": "course" } ], "failures": [], "summary": { "caseCount": 5, "successCount": 5, "crashCount": 0, "crashRate": 0, "averageWer": 0.0159, "averageCer": 0.0065, "averageInsertionRate": 0.001, "averageDeletionRate": 0.003, "averageSubstitutionRate": 0.0101, "averageDuplicateTokenRate": 0.1729, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 19118.4, "averageRealTimeFactor": 0.3016, "byScene": { "general": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0, "averageSubstitutionRate": 0, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 12983, "averageRealTimeFactor": 0.3759 }, "speech": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0, "averageSubstitutionRate": 0, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 21946, "averageRealTimeFactor": 0.3091 }, "dialogue": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0396, "averageCer": null, "averageInsertionRate": 0.005, "averageDeletionRate": 0.0149, "averageSubstitutionRate": 0.0198, "averageDuplicateTokenRate": 0.025, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 21858, "averageRealTimeFactor": 0.3193 }, "course": { "caseCount": 2, "successCount": 2, "crashCount": 0, "crashRate": 0, "averageWer": 0.024, "averageCer": 0.0065, "averageInsertionRate": 0, "averageDeletionRate": 0, "averageSubstitutionRate": 0.0152, "averageDuplicateTokenRate": 0.4197, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 19402.5, "averageRealTimeFactor": 0.2519 } } } } }, { "modelId": "parakeet", "modelArch": "parakeet", "batch": { "engine": "ml-engine-asr", "results": [ { "engine": "ml-engine-asr", "model": "parakeet", "caseId": "audio-benchmark", "segmentCount": 15, "errorMetricKind": "wer", "wer": 0, "cer": null, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0, "duplicateTokenRate": 0, "audioDurationMs": 34536, "preprocessElapsedMs": 0, "transcribeElapsedMs": 8601, "totalElapsedMs": 8601, "realTimeFactor": 0.2491, "sceneHint": "general" }, { "engine": "ml-engine-asr", "model": "parakeet", "caseId": "en-long-sentence-pause", "segmentCount": 33, "errorMetricKind": "wer", "wer": 0, "cer": null, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0, "duplicateTokenRate": 0, "audioDurationMs": 70992, "preprocessElapsedMs": 0, "transcribeElapsedMs": 10690, "totalElapsedMs": 10690, "realTimeFactor": 0.1506, "sceneHint": "speech" }, { "engine": "ml-engine-asr", "model": "parakeet", "caseId": "en-oral-correction", "segmentCount": 35, "errorMetricKind": "wer", "wer": 0.0396, "cer": null, "insertionRate": 0, "deletionRate": 0.0248, "substitutionRate": 0.0149, "duplicateTokenRate": 0.0254, "audioDurationMs": 68448, "preprocessElapsedMs": 0, "transcribeElapsedMs": 10535, "totalElapsedMs": 10535, "realTimeFactor": 0.1539, "sceneHint": "dialogue" }, { "engine": "ml-engine-asr", "model": "parakeet", "caseId": "en-tech-talk-podcast", "segmentCount": 26, "errorMetricKind": "wer", "wer": 0.024, "cer": null, "insertionRate": 0, "deletionRate": 0.018, "substitutionRate": 0.006, "duplicateTokenRate": 0.0061, "audioDurationMs": 79680, "preprocessElapsedMs": 0, "transcribeElapsedMs": 10130, "totalElapsedMs": 10130, "realTimeFactor": 0.1271, "sceneHint": "course" } ], "failures": [], "summary": { "caseCount": 4, "successCount": 4, "crashCount": 0, "crashRate": 0, "averageWer": 0.0159, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0.0107, "averageSubstitutionRate": 0.0052, "averageDuplicateTokenRate": 0.0079, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 9989, "averageRealTimeFactor": 0.1702, "byScene": { "general": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0, "averageSubstitutionRate": 0, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 8601, "averageRealTimeFactor": 0.2491 }, "speech": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0, "averageSubstitutionRate": 0, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 10690, "averageRealTimeFactor": 0.1506 }, "dialogue": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0396, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0.0248, "averageSubstitutionRate": 0.0149, "averageDuplicateTokenRate": 0.0254, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 10535, "averageRealTimeFactor": 0.1539 }, "course": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.024, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0.018, "averageSubstitutionRate": 0.006, "averageDuplicateTokenRate": 0.0061, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 10130, "averageRealTimeFactor": 0.1271 } } } } }, { "modelId": "sense-voice", "modelArch": "sense-voice", "batch": { "engine": "ml-engine-asr", "results": [ { "engine": "ml-engine-asr", "model": "sense-voice", "caseId": "audio-benchmark", "segmentCount": 15, "errorMetricKind": "wer", "wer": 0, "cer": null, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0, "duplicateTokenRate": 0, "audioDurationMs": 34536, "preprocessElapsedMs": 0, "transcribeElapsedMs": 2869, "totalElapsedMs": 2869, "realTimeFactor": 0.0831, "sceneHint": "general" }, { "engine": "ml-engine-asr", "model": "sense-voice", "caseId": "en-long-sentence-pause", "segmentCount": 33, "errorMetricKind": "wer", "wer": 0.0099, "cer": null, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0.0099, "duplicateTokenRate": 0, "audioDurationMs": 70992, "preprocessElapsedMs": 0, "transcribeElapsedMs": 4217, "totalElapsedMs": 4217, "realTimeFactor": 0.0594, "sceneHint": "speech" }, { "engine": "ml-engine-asr", "model": "sense-voice", "caseId": "en-oral-correction", "segmentCount": 35, "errorMetricKind": "wer", "wer": 0.0743, "cer": null, "insertionRate": 0.005, "deletionRate": 0.0297, "substitutionRate": 0.0396, "duplicateTokenRate": 0.0254, "audioDurationMs": 68448, "preprocessElapsedMs": 0, "transcribeElapsedMs": 4126, "totalElapsedMs": 4126, "realTimeFactor": 0.0603, "sceneHint": "dialogue" }, { "engine": "ml-engine-asr", "model": "sense-voice", "caseId": "en-tech-talk-podcast", "segmentCount": 26, "errorMetricKind": "wer", "wer": 0.0419, "cer": null, "insertionRate": 0, "deletionRate": 0.006, "substitutionRate": 0.0359, "duplicateTokenRate": 0.006, "audioDurationMs": 79680, "preprocessElapsedMs": 0, "transcribeElapsedMs": 3962, "totalElapsedMs": 3962, "realTimeFactor": 0.0497, "sceneHint": "course" }, { "engine": "ml-engine-asr", "model": "sense-voice", "caseId": "zh-tech-podcast", "segmentCount": 29, "errorMetricKind": "cer", "wer": null, "cer": 0.0161, "insertionRate": 0.0065, "deletionRate": 0, "substitutionRate": 0.0097, "duplicateTokenRate": 0.2857, "audioDurationMs": 74664, "preprocessElapsedMs": 0, "transcribeElapsedMs": 4078, "totalElapsedMs": 4078, "realTimeFactor": 0.0546, "sceneHint": "course" } ], "failures": [], "summary": { "caseCount": 5, "successCount": 5, "crashCount": 0, "crashRate": 0, "averageWer": 0.0315, "averageCer": 0.0161, "averageInsertionRate": 0.0023, "averageDeletionRate": 0.0071, "averageSubstitutionRate": 0.019, "averageDuplicateTokenRate": 0.0634, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 3850.4, "averageRealTimeFactor": 0.0614, "byScene": { "general": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0, "averageSubstitutionRate": 0, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 2869, "averageRealTimeFactor": 0.0831 }, "speech": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0099, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0, "averageSubstitutionRate": 0.0099, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 4217, "averageRealTimeFactor": 0.0594 }, "dialogue": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0743, "averageCer": null, "averageInsertionRate": 0.005, "averageDeletionRate": 0.0297, "averageSubstitutionRate": 0.0396, "averageDuplicateTokenRate": 0.0254, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 4126, "averageRealTimeFactor": 0.0603 }, "course": { "caseCount": 2, "successCount": 2, "crashCount": 0, "crashRate": 0, "averageWer": 0.0419, "averageCer": 0.0161, "averageInsertionRate": 0.0032, "averageDeletionRate": 0.003, "averageSubstitutionRate": 0.0228, "averageDuplicateTokenRate": 0.1459, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 4020, "averageRealTimeFactor": 0.0522 } } } } }, { "modelId": "fire-red-ctc", "modelArch": "fire-red-ctc", "batch": { "engine": "ml-engine-asr", "results": [ { "engine": "ml-engine-asr", "model": "fire-red-ctc", "caseId": "audio-benchmark", "segmentCount": 15, "errorMetricKind": "wer", "wer": 0.0753, "cer": null, "insertionRate": 0.0108, "deletionRate": 0, "substitutionRate": 0.0645, "duplicateTokenRate": 0, "audioDurationMs": 34536, "preprocessElapsedMs": 0, "transcribeElapsedMs": 10922, "totalElapsedMs": 10922, "realTimeFactor": 0.3162, "sceneHint": "general" }, { "engine": "ml-engine-asr", "model": "fire-red-ctc", "caseId": "en-long-sentence-pause", "segmentCount": 33, "errorMetricKind": "wer", "wer": 0.1576, "cer": null, "insertionRate": 0.0099, "deletionRate": 0.0049, "substitutionRate": 0.1429, "duplicateTokenRate": 0, "audioDurationMs": 70992, "preprocessElapsedMs": 0, "transcribeElapsedMs": 20089, "totalElapsedMs": 20089, "realTimeFactor": 0.283, "sceneHint": "speech" }, { "engine": "ml-engine-asr", "model": "fire-red-ctc", "caseId": "en-oral-correction", "segmentCount": 35, "errorMetricKind": "wer", "wer": 0.1089, "cer": null, "insertionRate": 0.0099, "deletionRate": 0.0198, "substitutionRate": 0.0792, "duplicateTokenRate": 0.025, "audioDurationMs": 68448, "preprocessElapsedMs": 0, "transcribeElapsedMs": 19331, "totalElapsedMs": 19331, "realTimeFactor": 0.2824, "sceneHint": "dialogue" }, { "engine": "ml-engine-asr", "model": "fire-red-ctc", "caseId": "en-tech-talk-podcast", "segmentCount": 26, "errorMetricKind": "wer", "wer": 0.1317, "cer": null, "insertionRate": 0.018, "deletionRate": 0.012, "substitutionRate": 0.1018, "duplicateTokenRate": 0.006, "audioDurationMs": 79680, "preprocessElapsedMs": 0, "transcribeElapsedMs": 19086, "totalElapsedMs": 19086, "realTimeFactor": 0.2395, "sceneHint": "course" }, { "engine": "ml-engine-asr", "model": "fire-red-ctc", "caseId": "zh-tech-podcast", "segmentCount": 29, "errorMetricKind": "cer", "wer": null, "cer": 0.0323, "insertionRate": 0.0065, "deletionRate": 0.0129, "substitutionRate": 0.0129, "duplicateTokenRate": 0.4, "audioDurationMs": 74664, "preprocessElapsedMs": 0, "transcribeElapsedMs": 19689, "totalElapsedMs": 19689, "realTimeFactor": 0.2637, "sceneHint": "course" } ], "failures": [], "summary": { "caseCount": 5, "successCount": 5, "crashCount": 0, "crashRate": 0, "averageWer": 0.1184, "averageCer": 0.0323, "averageInsertionRate": 0.011, "averageDeletionRate": 0.0099, "averageSubstitutionRate": 0.0803, "averageDuplicateTokenRate": 0.0862, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 17823.4, "averageRealTimeFactor": 0.277, "byScene": { "general": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0753, "averageCer": null, "averageInsertionRate": 0.0108, "averageDeletionRate": 0, "averageSubstitutionRate": 0.0645, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 10922, "averageRealTimeFactor": 0.3162 }, "speech": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.1576, "averageCer": null, "averageInsertionRate": 0.0099, "averageDeletionRate": 0.0049, "averageSubstitutionRate": 0.1429, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 20089, "averageRealTimeFactor": 0.283 }, "dialogue": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.1089, "averageCer": null, "averageInsertionRate": 0.0099, "averageDeletionRate": 0.0198, "averageSubstitutionRate": 0.0792, "averageDuplicateTokenRate": 0.025, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 19331, "averageRealTimeFactor": 0.2824 }, "course": { "caseCount": 2, "successCount": 2, "crashCount": 0, "crashRate": 0, "averageWer": 0.1317, "averageCer": 0.0323, "averageInsertionRate": 0.0122, "averageDeletionRate": 0.0124, "averageSubstitutionRate": 0.0573, "averageDuplicateTokenRate": 0.203, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 19387.5, "averageRealTimeFactor": 0.2516 } } } } }, { "modelId": "moonshine", "modelArch": "moonshine", "batch": { "engine": "ml-engine-asr", "results": [ { "engine": "ml-engine-asr", "model": "moonshine", "caseId": "audio-benchmark", "segmentCount": 15, "errorMetricKind": "wer", "wer": 0, "cer": null, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0, "duplicateTokenRate": 0, "audioDurationMs": 34536, "preprocessElapsedMs": 0, "transcribeElapsedMs": 1407, "totalElapsedMs": 1407, "realTimeFactor": 0.0407, "sceneHint": "general" }, { "engine": "ml-engine-asr", "model": "moonshine", "caseId": "en-long-sentence-pause", "segmentCount": 33, "errorMetricKind": "wer", "wer": 0.0542, "cer": null, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0.0542, "duplicateTokenRate": 0, "audioDurationMs": 70992, "preprocessElapsedMs": 0, "transcribeElapsedMs": 2027, "totalElapsedMs": 2027, "realTimeFactor": 0.0286, "sceneHint": "speech" }, { "engine": "ml-engine-asr", "model": "moonshine", "caseId": "en-oral-correction", "segmentCount": 35, "errorMetricKind": "wer", "wer": 0.0594, "cer": null, "insertionRate": 0.005, "deletionRate": 0.0149, "substitutionRate": 0.0396, "duplicateTokenRate": 0.025, "audioDurationMs": 68448, "preprocessElapsedMs": 0, "transcribeElapsedMs": 2072, "totalElapsedMs": 2072, "realTimeFactor": 0.0303, "sceneHint": "dialogue" }, { "engine": "ml-engine-asr", "model": "moonshine", "caseId": "en-tech-talk-podcast", "segmentCount": 26, "errorMetricKind": "wer", "wer": 0.0479, "cer": null, "insertionRate": 0.024, "deletionRate": 0.006, "substitutionRate": 0.018, "duplicateTokenRate": 0.0059, "audioDurationMs": 79680, "preprocessElapsedMs": 0, "transcribeElapsedMs": 2016, "totalElapsedMs": 2016, "realTimeFactor": 0.0253, "sceneHint": "course" } ], "failures": [], "summary": { "caseCount": 4, "successCount": 4, "crashCount": 0, "crashRate": 0, "averageWer": 0.0404, "averageCer": null, "averageInsertionRate": 0.0073, "averageDeletionRate": 0.0052, "averageSubstitutionRate": 0.0279, "averageDuplicateTokenRate": 0.0077, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 1880.5, "averageRealTimeFactor": 0.0312, "byScene": { "general": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0, "averageSubstitutionRate": 0, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 1407, "averageRealTimeFactor": 0.0407 }, "speech": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0542, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0, "averageSubstitutionRate": 0.0542, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 2027, "averageRealTimeFactor": 0.0286 }, "dialogue": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0594, "averageCer": null, "averageInsertionRate": 0.005, "averageDeletionRate": 0.0149, "averageSubstitutionRate": 0.0396, "averageDuplicateTokenRate": 0.025, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 2072, "averageRealTimeFactor": 0.0303 }, "course": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0479, "averageCer": null, "averageInsertionRate": 0.024, "averageDeletionRate": 0.006, "averageSubstitutionRate": 0.018, "averageDuplicateTokenRate": 0.0059, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 2016, "averageRealTimeFactor": 0.0253 } } } } }, { "modelId": "cohere-transcribe-multilingual", "modelArch": "cohere", "batch": { "engine": "ml-engine-asr", "results": [ { "engine": "ml-engine-asr", "model": "cohere", "caseId": "audio-benchmark", "segmentCount": 15, "errorMetricKind": "wer", "wer": 0.0108, "cer": null, "insertionRate": 0, "deletionRate": 0.0108, "substitutionRate": 0, "duplicateTokenRate": 0, "audioDurationMs": 34536, "preprocessElapsedMs": 0, "transcribeElapsedMs": 16310, "totalElapsedMs": 16310, "realTimeFactor": 0.4723, "sceneHint": "general" }, { "engine": "ml-engine-asr", "model": "cohere", "caseId": "en-long-sentence-pause", "segmentCount": 33, "errorMetricKind": "wer", "wer": 0.0049, "cer": null, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0.0049, "duplicateTokenRate": 0, "audioDurationMs": 70992, "preprocessElapsedMs": 0, "transcribeElapsedMs": 24976, "totalElapsedMs": 24976, "realTimeFactor": 0.3518, "sceneHint": "speech" }, { "engine": "ml-engine-asr", "model": "cohere", "caseId": "en-oral-correction", "segmentCount": 35, "errorMetricKind": "wer", "wer": 0.0495, "cer": null, "insertionRate": 0, "deletionRate": 0.0149, "substitutionRate": 0.0347, "duplicateTokenRate": 0.0251, "audioDurationMs": 68448, "preprocessElapsedMs": 0, "transcribeElapsedMs": 24552, "totalElapsedMs": 24552, "realTimeFactor": 0.3587, "sceneHint": "dialogue" }, { "engine": "ml-engine-asr", "model": "cohere", "caseId": "en-tech-talk-podcast", "segmentCount": 26, "errorMetricKind": "wer", "wer": 0.006, "cer": null, "insertionRate": 0, "deletionRate": 0, "substitutionRate": 0.006, "duplicateTokenRate": 0.006, "audioDurationMs": 79680, "preprocessElapsedMs": 0, "transcribeElapsedMs": 23163, "totalElapsedMs": 23163, "realTimeFactor": 0.2907, "sceneHint": "course" }, { "engine": "ml-engine-asr", "model": "cohere", "caseId": "zh-tech-podcast", "segmentCount": 29, "errorMetricKind": "cer", "wer": null, "cer": 0.1032, "insertionRate": 0, "deletionRate": 0.0903, "substitutionRate": 0.0129, "duplicateTokenRate": 0.5, "audioDurationMs": 74664, "preprocessElapsedMs": 0, "transcribeElapsedMs": 26183, "totalElapsedMs": 26183, "realTimeFactor": 0.3507, "sceneHint": "course" } ], "failures": [], "summary": { "caseCount": 5, "successCount": 5, "crashCount": 0, "crashRate": 0, "averageWer": 0.0178, "averageCer": 0.1032, "averageInsertionRate": 0, "averageDeletionRate": 0.0232, "averageSubstitutionRate": 0.0117, "averageDuplicateTokenRate": 0.1062, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 23036.8, "averageRealTimeFactor": 0.3648, "byScene": { "general": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0108, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0.0108, "averageSubstitutionRate": 0, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 16310, "averageRealTimeFactor": 0.4723 }, "speech": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0049, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0, "averageSubstitutionRate": 0.0049, "averageDuplicateTokenRate": 0, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 24976, "averageRealTimeFactor": 0.3518 }, "dialogue": { "caseCount": 1, "successCount": 1, "crashCount": 0, "crashRate": 0, "averageWer": 0.0495, "averageCer": null, "averageInsertionRate": 0, "averageDeletionRate": 0.0149, "averageSubstitutionRate": 0.0347, "averageDuplicateTokenRate": 0.0251, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 24552, "averageRealTimeFactor": 0.3587 }, "course": { "caseCount": 2, "successCount": 2, "crashCount": 0, "crashRate": 0, "averageWer": 0.006, "averageCer": 0.1032, "averageInsertionRate": 0, "averageDeletionRate": 0.0452, "averageSubstitutionRate": 0.0095, "averageDuplicateTokenRate": 0.253, "averagePreprocessElapsedMs": 0, "averageTotalElapsedMs": 24673, "averageRealTimeFactor": 0.3207 } } } } } ], "skipped": [ { "modelId": "en-zipformer-onnx", "caseId": "zh-tech-podcast", "reason": "language_not_supported" }, { "modelId": "parakeet", "caseId": "zh-tech-podcast", "reason": "language_not_supported" }, { "modelId": "moonshine", "caseId": "zh-tech-podcast", "reason": "language_not_supported" } ] }