{
  "id": "human-conversation",
  "title": "Two speakers: booking and recording equipment",
  "kind": "human",
  "meetingId": "EN2009d",
  "durationSec": 96.57,
  "sourceStartSec": 4926.96,
  "sourceEndSec": 5023.53,
  "sourceUrl": "https://groups.inf.ed.ac.uk/ami/AMICorpusMirror/amicorpus/HeadsetAudio/EN2009d.Mix-Headset.wav",
  "sourcePage": "https://groups.inf.ed.ac.uk/ami/corpus/",
  "sourceDescriptionPage": "https://groups.inf.ed.ac.uk/ami/corpus/nonscenariomeetings.shtml",
  "license": "CC BY 4.0",
  "licenseUrl": "https://creativecommons.org/licenses/by/4.0/",
  "licenseEvidence": "https://groups.inf.ed.ac.uk/ami/corpus/license.shtml",
  "creator": "The AMI Consortium",
  "copyright": "Copyright (C) 2006, The AMI Consortium",
  "attribution": "AMI Meeting Corpus, The AMI Consortium / University of Edinburgh; CC BY 4.0. Contiguous excerpt, original headset mix. No endorsement implied.",
  "modifications": "Single contiguous excerpt from the original 16 kHz mono PCM headset mix. WAV container rebuilt; no stitching, synthetic speech, voice isolation, denoising, silence removal, speed changes or word edits.",
  "audioUrl": "/samples/whisperx/human-conversation/recording.wav",
  "referenceUrl": "/samples/whisperx/human-conversation/reference.words.json",
  "sourceSpeakerIds": [
    "A",
    "B"
  ],
  "sourceMeetingHasMoreParticipants": true,
  "audioSha256": "e42f0edd3c5403f9607636d366aa584d8c4a632e104691f406ddf9fb7d2da588",
  "annotationArchiveSha256": "b56e5babb2496b8795deeeda7e71178d7fbc9963f94276cf2a3f4b56ebbc9f9d",
  "downloadedAt": "2026-10-15T07:42:10.086567+00:00",
  "contentRange": "bytes 157662764-160753003/170379992",
  "sourceFullAudioHash": null,
  "reviewStatus": "Completed: reference-based comparison; no listening audit",
  "processedAt": "2026-10-15T07:44:06.508Z",
  "predictionId": "ngcd0fcd0hrne0d17t292erjqm",
  "engine": {
    "name": "WhisperX",
    "provider": "replicate",
    "model": "large-v3",
    "version": "655845d6190ef70573c669245f245892cd039df4b880a1e3a65852c09252f5cc"
  },
  "configuration": {
    "language": "en",
    "task": "transcribe",
    "align_output": true,
    "diarization": true,
    "batch_size": 32,
    "min_speakers": "unset (automatic)",
    "max_speakers": "unset (automatic)",
    "hfModel": "pyannote/speaker-diarization-community-1",
    "cancelAfter": "300s"
  },
  "executionPath": "direct provider call through unchanged production submitPrediction/buildProviderInput/normalizer; test-only Cancel-After header; no app job, credit or DB changes",
  "metrics": {
    "predict_time": 7.083176214,
    "total_time": 7.094790251
  },
  "predictionCreatedAt": "2026-10-15T07:43:56.676Z",
  "predictionStartedAt": "2026-10-15T07:43:56.687614037Z",
  "predictionCompletedAt": "2026-10-15T07:44:03.770790251Z",
  "budget": {
    "publishedHardware": "A100 80GB",
    "publishedRateUSDPerSecond": 0.0014,
    "estimatedUSD": 0.0099164466996,
    "actualBilledUSD": null
  },
  "outputFiles": {
    "provider": "/samples/whisperx/human-conversation/provider.json",
    "json": "/samples/whisperx/human-conversation/transcript.json",
    "txt": "/samples/whisperx/human-conversation/transcript.txt",
    "srt": "/samples/whisperx/human-conversation/transcript.srt",
    "vtt": "/samples/whisperx/human-conversation/transcript.vtt"
  },
  "review": {
    "summary": "The model detected two speakers, but speaker assignments and short replies still need review.",
    "observations": [
      "The first model segment (“Because, you know…booking system”) corresponds to source speaker A, while the next (“Well, we think…”) corresponds to B. Both are labeled SPEAKER_00, so this boundary does not preserve the two source speakers.",
      "The 38.30–43.35s account-related statements are from reference speaker B, but the model uses SPEAKER_01, also used for speaker A’s earlier question. The labels are not consistently assigned.",
      "The reference’s “the two key holders are here” is not present in the model transcript, and opening/overlapping short replies are omitted.",
      "The person-name reference “Marloes” appears as “Marlos”; “Caroline” is retained and J_P_ is rendered JP.",
      "At 58.04–60.92s “Because then they learn the first time” is assigned SPEAKER_00 even though the corresponding source words are annotated as A."
    ],
    "method": "Reference-based comparison against the AMI manual word annotations v1.6.2, including speaker IDs and original source times. Not a listening audit.",
    "reviewedAt": "2026-10-15T07:45:32.764850+00:00",
    "notManuallyListened": true,
    "measured": {
      "durationSec": 96.57,
      "modelSegments": 25,
      "modelWordEntries": 233,
      "wordEntriesWithTiming": 233,
      "missingWordTiming": 0,
      "modelSpeakerCount": 2,
      "referenceSpeakerCount": 2,
      "referenceSpeakerIds": [
        "A",
        "B"
      ],
      "referenceLexicalEntries": 307,
      "timingStatusCounts": {
        "unknown": 233
      }
    },
    "limitations": [
      "These are two short, selected excerpts from real non-scenario multi-person meetings, using the official mixed headset recording. They are not privately recorded one-person/two-person sessions.",
      "The reference contains false starts, fillers and overlapping speech. Word-count differences are not an error rate, and no WER/DER or general accuracy claim is made.",
      "Timestamps are model output, not independently listened-to alignment measurements. All normalized timing statuses remain unknown, even where numeric times exist.",
      "No output text, speaker label or timestamp was manually corrected. Original provider output and normalized original exports are preserved.",
      "This direct-provider test uses the unchanged production input adapter, pinned model, normalizer and exporters. It does not test application login, credits, job database, webhook handling or browser processing."
    ]
  },
  "reviewUrl": "/samples/whisperx/human-conversation/review.json",
  "notesUrl": "/samples/whisperx/human-conversation/review.md",
  "detectedSpeakerCount": 2,
  "wordCount": 233,
  "segmentCount": 25,
  "sourceLabel": "Human speech · AMI non-scenario meeting · CC BY 4.0"
}