{
  "@context": "https://schema.org",
  "@type": "Dataset",
  "aggregate": {
    "cer": 0.13826185101580135,
    "character_errors": 245,
    "character_reference_units": 1772,
    "cost_samples": 0,
    "critical_term_matches": 51,
    "critical_term_recall": 0.85,
    "critical_term_targets": 60,
    "number_matches": 0,
    "number_recall": 1.0,
    "number_targets": 0,
    "processing_seconds": 11.231269,
    "processing_seconds_mean": 0.56156345,
    "real_time_factor": 0.0612175,
    "speaker_attribution_samples": 0,
    "wer": 0.21261682242990654,
    "word_errors": 91,
    "word_reference_units": 428
  },
  "audio_duration_seconds": 183.465,
  "benchmark_id": "longscribe-transcription-benchmark-v1",
  "contentUrl": "https://longscribe.com/data/longscribe-transcription-benchmark-v1.json",
  "creator": {
    "@type": "Organization",
    "name": "LongScribe",
    "url": "https://longscribe.com/"
  },
  "dataset_frozen_at": "2026-09-11T15:12:00Z",
  "dataset_manifest_sha256": "d232f4a4dde5dfc8dcd89156b1bacc702ff2441cd729c77e3818b2679881e2cb",
  "dateModified": "2026-09-11",
  "description": "Frozen 20-sample benchmark of LongScribe's production-primary ASR path, with aggregate and per-slice WER, CER, critical-term recall and timing.",
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "limitations": [
    "Only LongScribe was run on this frozen set, so these results are not a cross-vendor accuracy ranking.",
    "The frozen benchmark contains short clips; it does not measure multi-hour end-to-end job latency.",
    "Speaker attribution was not measured in this run.",
    "The frozen references contain zero numeric targets, so the reported 1.0 number-recall field is not evidence of numeric accuracy.",
    "No provider cost sample was recorded in this run."
  ],
  "name": "LongScribe production-primary transcription benchmark v1",
  "normalization_version": "asr-text-v1",
  "provider": {
    "name": "LongScribe",
    "settings": {
      "backend": "groq",
      "execution": "sequential isolated benchmark; no production job or quota entry",
      "fallback_used": false,
      "language": "auto",
      "live_app_main_sha256": "9a223580afbf05ddc33ffaad4746d0188d364fa1b80037b5818dc9b4b616e894",
      "model": "whisper-large-v3-turbo",
      "pipeline": "current-main production-primary ASR contract",
      "preprocessing": "ffmpeg mono 16 kHz PCM s16le WAV",
      "quality_retry_policy": "same backend once only for suspicious sparse output",
      "response_format": "verbose_json",
      "source_main_sha": "94f20341952d2365ad05919f6ea042d1b5c938a3",
      "temperature": 0.1,
      "timestamp_granularity": "segment"
    },
    "tested_at": "2026-09-11T16:12:07Z"
  },
  "run_manifest_sha256": "e10404d5dd401005f7688848c715e7e206b5cdd423f59ee8e42578d6ed5d59b0",
  "sample_count": 20,
  "slices": [
    {
      "audio_duration_seconds": 50.175,
      "cer": 0.0,
      "character_errors": 0,
      "character_reference_units": 587,
      "critical_term_matches": 15,
      "critical_term_recall": 1.0,
      "critical_term_targets": 15,
      "number_matches": 0,
      "number_recall": 1.0,
      "number_targets": 0,
      "processing_seconds": 3.630529,
      "real_time_factor": 0.0723573293472845,
      "sample_count": 5,
      "slice": "clean_single_speaker",
      "wer": 0.016,
      "word_errors": 2,
      "word_reference_units": 125
    },
    {
      "audio_duration_seconds": 59.68,
      "cer": 0.28255528255528256,
      "character_errors": 115,
      "character_reference_units": 407,
      "critical_term_matches": 10,
      "critical_term_recall": 0.6666666666666666,
      "critical_term_targets": 15,
      "number_matches": 0,
      "number_recall": 1.0,
      "number_targets": 0,
      "processing_seconds": 3.247762,
      "real_time_factor": 0.05441960455764075,
      "sample_count": 5,
      "slice": "multi_speaker_overlap",
      "wer": 0.36134453781512604,
      "word_errors": 43,
      "word_reference_units": 119
    },
    {
      "audio_duration_seconds": 36.75,
      "cer": 0.22302158273381295,
      "character_errors": 93,
      "character_reference_units": 417,
      "critical_term_matches": 12,
      "critical_term_recall": 0.8,
      "critical_term_targets": 15,
      "number_matches": 0,
      "number_recall": 1.0,
      "number_targets": 0,
      "processing_seconds": 2.052515,
      "real_time_factor": 0.05585074829931973,
      "sample_count": 5,
      "slice": "noisy_far_field",
      "wer": 0.3010752688172043,
      "word_errors": 28,
      "word_reference_units": 93
    },
    {
      "audio_duration_seconds": 36.86,
      "cer": 0.10249307479224377,
      "character_errors": 37,
      "character_reference_units": 361,
      "critical_term_matches": 14,
      "critical_term_recall": 0.9333333333333333,
      "critical_term_targets": 15,
      "number_matches": 0,
      "number_recall": 1.0,
      "number_targets": 0,
      "processing_seconds": 2.300463,
      "real_time_factor": 0.06241082474226805,
      "sample_count": 5,
      "slice": "non_native_single_speaker",
      "wer": 0.1978021978021978,
      "word_errors": 18,
      "word_reference_units": 91
    }
  ],
  "sources": [
    {
      "license": "CC BY 4.0",
      "name": "AMI Meeting Corpus",
      "url": "https://groups.inf.ed.ac.uk/ami/AMICorpusMirror/amicorpus/ES2004a/audio/ES2004a.Array1-01.wav"
    },
    {
      "license": "CC BY 4.0",
      "name": "AMI Meeting Corpus",
      "url": "https://groups.inf.ed.ac.uk/ami/AMICorpusMirror/amicorpus/ES2004a/audio/ES2004a.Mix-Headset.wav"
    },
    {
      "license": "CC BY 4.0",
      "name": "AMI Meeting Corpus",
      "url": "https://groups.inf.ed.ac.uk/ami/AMICorpusMirror/amicorpus/IS1000a/audio/IS1000a.Array1-01.wav"
    },
    {
      "license": "CC BY 4.0",
      "name": "AMI Meeting Corpus",
      "url": "https://groups.inf.ed.ac.uk/ami/AMICorpusMirror/amicorpus/IS1000a/audio/IS1000a.Mix-Headset.wav"
    },
    {
      "license": "CC BY 4.0",
      "name": "LibriSpeech ASR corpus",
      "url": "https://www.openslr.org/resources/12/test-clean.tar.gz"
    }
  ],
  "tested_at": "2026-09-11T16:12:07Z",
  "url": "https://longscribe.com/best-ai-transcription-software"
}