{
  "schema": "transcribe-golden-manifest-v1",
  "family": "voxtral",
  "variant": "voxtral-mini-3b-2507",
  "source_model": {
    "hf_repo": "mistralai/Voxtral-Mini-3B-2507",
    "hf_revision": "3060fe34b35ba5d44202ce9ff3c097642914f8f3"
  },
  "reference": {
    "kind": "transformers",
    "source": "https://github.com/huggingface/transformers",
    "revision": "v4.57.6",
    "entrypoint": "scripts/dump_reference_voxtral_transformers.py",
    "dump_args": ["--language", "en"]
  },
  "expected_dtype": "bfloat16",
  "dtype_source": "config",
  "frontend": {
    "sample_rate": 16000,
    "n_mels": 128,
    "hop_length": 160,
    "fft_size": 400,
    "win_length": 400,
    "window": "hann_periodic",
    "normalization": "per_utterance",
    "preemphasis": null,
    "dither": 0.0
  },
  "tokenizer_summary": {
    "type": "bpe",
    "vocab_size": 131072,
    "special_tokens": {
      "bos": 1,
      "eos": 2,
      "pad": 11,
      "audio": 24
    }
  },
  "capabilities": {
    "languages": ["en", "fr", "de", "es", "it", "pt", "nl", "hi"],
    "language_detection": true,
    "translation": true,
    "translation_target_languages": ["en", "fr", "de", "es", "it", "pt", "nl", "hi"],
    "timestamps": ["none"],
    "streaming": false,
    "speaker_diarization": false
  },
  "tolerance_file": "tests/tolerances/voxtral.json",
  "cases": ["jfk"]
}
