{
  "software_date": "2026-09-29",
  "dataset": "Google FLEURS",
  "revision": "70bb2e84b976b7e960aa89f1c648e09c59f894dd",
  "seed": 20260929,
  "originals": 144,
  "diagnostics": 4,
  "selection": "SHA-256 rank(seed|config|sentence_id|filename); first twelve distinct references and filenames per language; no result-based reselection",
  "public_examples": [
    "de-01",
    "de-06",
    "en-01",
    "en-04",
    "es-01",
    "es-04",
    "fr-01",
    "fr-07",
    "pt-01",
    "pt-10",
    "it-01",
    "it-12",
    "ja-01",
    "ja-04",
    "ko-01",
    "ko-05",
    "tr-01",
    "tr-08",
    "ru-01",
    "ru-11",
    "ar-01",
    "ar-02",
    "zh-01",
    "zh-03",
    "diag-silence",
    "diag-ja-pause",
    "diag-ja-noise",
    "diag-de-noise"
  ],
  "public_selection": "original rank 1 plus highest CER per language; rank 2 if same; failures plus four separate diagnostics",
  "settings": {
    "quality": "middle",
    "language": "auto",
    "model": "large-v3",
    "beam_size": 3,
    "best_of": 3,
    "cli": "Standalone Faster-Whisper-XXL r245.2",
    "faster_whisper": "1.1.1",
    "translation": false,
    "optional_audio_enhancement": false,
    "speaker_diarization": false,
    "pcm": "16-bit 16kHz mono",
    "vad": "silero_v4_fw",
    "vad_threshold": 0.32,
    "min_speech_ms": 100,
    "min_silence_ms": 700,
    "pad_ms": 900,
    "condition_on_previous_text": false,
    "cjk": "content-based analysis and simple repair active; no Mandarin robust/LLM run"
  },
  "coverage": {
    "isolated_complete_product_processing": 148,
    "regular_upload_download": 1,
    "regular_audio_language": "ja",
    "human_listening_approval": false
  },
  "normalization": "Unicode NFC, lowercase (Turkish I->ı, İ->i), punctuation as boundaries; numeric signs/percent and separators between digits retained; no number-word conversion, no removal of negations/names/Arabic diacritics or script folding. WER whitespace units; JA/ZH no WER; CER codepoints without spaces. Rates: summed S+D+I / summed N, not mean percentages. Silence: no rate.",
  "speaker_limit": "No reliable speaker IDs; gender metadata is not identity. Training overlap not excluded.",
  "history": {
    "silence_placeholder_fixed": "2026-09-30",
    "historical_files_unchanged": true,
    "concluding_phrase_detection": "unpublished prototype"
  }
}