{
  "model_layout": "offline_encoder",
  "blank_idx": 13087,
  "vocab_size": 13087,
  "decoder_hidden": 640,
  "decoder_layers": 2,
  "sample_rate": 16000,
  "mel_features": 128,
  "max_mel_frames": 1501,
  "max_audio_samples": 240000,
  "num_prompts": 128,
  "runtime_prompt": true,
  "target_lang": "pt",
  "prompt_dictionary": {
    "af-ZA": 54,
    "am-ET": 49,
    "ar": 7,
    "ar-AR": 7,
    "auto": 101,
    "ay-BO": 81,
    "az-AZ": 66,
    "bg": 30,
    "bg-BG": 30,
    "bn-IN": 36,
    "cs": 22,
    "cs-CZ": 22,
    "da": 25,
    "da-DK": 25,
    "de": 9,
    "de-DE": 9,
    "el": 21,
    "el-GR": 21,
    "en": 0,
    "en-GB": 1,
    "en-US": 0,
    "enGB": 1,
    "es": 3,
    "es-ES": 2,
    "es-US": 3,
    "esES": 2,
    "et": 60,
    "et-EE": 60,
    "fa-IR": 38,
    "fi": 26,
    "fi-FI": 26,
    "fr": 8,
    "fr-CA": 100,
    "fr-FR": 8,
    "gn-PY": 82,
    "gu-IN": 42,
    "ha-NG": 50,
    "haw-US": 97,
    "he-IL": 64,
    "hi": 6,
    "hi-HI": 6,
    "hi-IN": 6,
    "hr": 29,
    "hr-HR": 29,
    "hu": 23,
    "hu-HU": 23,
    "hy-AM": 68,
    "id-ID": 34,
    "ig-NG": 53,
    "it": 15,
    "it-IT": 15,
    "ja-JA": 10,
    "ja-JP": 10,
    "ka-GE": 67,
    "km-KH": 47,
    "kn-IN": 43,
    "ko": 14,
    "ko-KO": 14,
    "ko-KR": 14,
    "ku-TR": 65,
    "ky-KG": 71,
    "ln-CD": 58,
    "lt": 31,
    "lt-LT": 31,
    "lv": 61,
    "lv-LV": 61,
    "mi-NZ": 96,
    "ml-IN": 44,
    "mr-IN": 41,
    "ms-MY": 35,
    "mt-MT": 102,
    "nah-MX": 83,
    "nb": 103,
    "nb-NO": 103,
    "ne-NP": 46,
    "nl": 16,
    "nl-NL": 16,
    "nn": 104,
    "nn-NO": 104,
    "no": 27,
    "no-NO": 27,
    "ny-MW": 57,
    "or-KE": 59,
    "pl": 17,
    "pl-PL": 17,
    "pt": 13,
    "pt-BR": 12,
    "pt-PT": 13,
    "qu-PE": 80,
    "ro": 20,
    "ro-RO": 20,
    "ru": 11,
    "ru-RU": 11,
    "rw-RW": 55,
    "si-LK": 45,
    "sk": 28,
    "sk-SK": 28,
    "sl": 62,
    "sl-SI": 62,
    "sm-WS": 98,
    "so-SO": 56,
    "sv": 24,
    "sv-SE": 24,
    "sw-KE": 48,
    "ta-IN": 39,
    "te-IN": 40,
    "tg-TJ": 70,
    "th-TH": 32,
    "to-TO": 99,
    "tr": 18,
    "tr-TR": 18,
    "uk": 19,
    "uk-UA": 19,
    "ur-PK": 37,
    "uz-UZ": 69,
    "vi-VN": 33,
    "yo-NG": 52,
    "zh-CN": 4,
    "zh-TW": 5,
    "zh-ZH": 4,
    "zu-ZA": 51
  },
  "tokenizer_model": "tokenizer.model",
  "encoder_dim": 1024,
  "coreml": {
    "components": {
      "preprocessor": "preprocessor.mlpackage",
      "encoder": "encoder.mlpackage",
      "decoder": "decoder.mlpackage",
      "joint": "joint.mlpackage",
      "joint_decision": "joint_decision.mlpackage"
    },
    "compute_precision": "FLOAT32",
    "encoder_compute_units": "CPU_ONLY",
    "preprocessor_audio_flexible": true
  },
  "shapes": {
    "processed_signal": [
      1,
      128,
      1501
    ],
    "encoded": [
      1,
      1024,
      189
    ],
    "audio_signal": [
      1,
      240000
    ],
    "cache_last_channel": [
      24,
      1,
      56,
      1024
    ],
    "cache_last_time": [
      24,
      1,
      1024,
      8
    ]
  },
  "offline_encoder_parity_max_abs_diff": 1.811981201171875e-05,
  "quantization": {
    "encoder": {
      "method": "palettize_6bit",
      "baseline_mb": 2280.7,
      "quantized_mb": 429.4,
      "ratio": 5.31
    },
    "joint": {
      "method": "palettize_6bit",
      "baseline_mb": 36.1,
      "quantized_mb": 6.8,
      "ratio": 5.29
    },
    "nbits": 6,
    "method": "palettize",
    "decoder": {
      "method": "none_fp32",
      "note": "kept FP32 (57MB) by request"
    }
  }
}