diodiogod-TTS-Audio-Suite/utils/audio_cpp/model_specs/higgs_audio_stt.json

{
  "schema_version": 1,
  "family": "higgs_audio_stt",
  "display_name": "Higgs Audio v3 STT",
  "description": "Boson AI English speech-to-text model combining a Whisper Large v3 speech encoder with a Qwen decoder for robust ASR.",
  "category": "asr",
  "status": "supported",
  "tasks": [
    "asr"
  ],
  "modes": [
    "offline",
    "streaming"
  ],
  "languages": [
    "en"
  ],
  "capabilities": {},
  "dependencies": [],
  "options": {
    "request": [
      {
        "name": "language",
        "type": "string",
        "description": "Transcript language code metadata; English is used when omitted.",
        "required": false
      },
      {
        "name": "max_tokens",
        "type": "int",
        "description": "Maximum generated transcript tokens; default 1024.",
        "required": false,
        "min": 1,
        "default": 1024
      },
      {
        "name": "enable_thinking",
        "type": "bool",
        "description": "Enable the model thinking prompt; default true.",
        "required": false,
        "default": true
      },
      {
        "name": "audio_chunk_mode",
        "type": "enum",
        "description": "Audio chunking mode; default auto uses fixed chunks.",
        "values": [
          "auto",
          "fixed",
          "none"
        ],
        "required": false,
        "default": "auto"
      },
      {
        "name": "audio_chunk_duration_sec",
        "type": "float",
        "description": "Fixed audio chunk duration in seconds; must be positive when set; default 4.",
        "required": false,
        "min": 0.0,
        "default": 4.0
      }
    ],
    "session": [
      {
        "name": "weight_type",
        "type": "enum",
        "description": "Shared text decoder weight storage type; default native.",
        "preset": "weight_type_full",
        "required": false,
        "default": "native"
      },
      {
        "name": "audio_encoder_weight_type",
        "type": "enum",
        "description": "Audio encoder convolution weight storage type; default native.",
        "preset": "weight_type_conv",
        "required": false,
        "default": "native"
      },
      {
        "name": "text_decoder_weight_type",
        "type": "enum",
        "description": "Text decoder matmul weight storage type; defaults to weight_type when set, otherwise native.",
        "preset": "weight_type_full",
        "required": false
      },
      {
        "name": "audio_encoder_graph_arena_mb",
        "type": "int",
        "description": "Audio encoder graph arena size in MiB; default 512.",
        "required": false,
        "min": 0,
        "default": 512
      },
      {
        "name": "text_decoder_prefill_graph_arena_mb",
        "type": "int",
        "description": "Text decoder prefill graph arena size in MiB; default 512.",
        "required": false,
        "min": 0,
        "default": 512
      },
      {
        "name": "text_decoder_decode_graph_arena_mb",
        "type": "int",
        "description": "Text decoder cached-step graph arena size in MiB; default 256.",
        "required": false,
        "min": 0,
        "default": 256
      },
      {
        "name": "text_decoder_weight_context_mb",
        "type": "int",
        "description": "Text decoder weight context arena size in MiB; default 4096.",
        "required": false,
        "min": 0,
        "default": 4096
      }
    ],
    "load": []
  },
  "runtime": {
    "tags": [
      "gguf",
      "stream"
    ]
  },
  "ui": {
    "recommended_package": "higgs_audio_stt_q8_0",
    "tags": [
      "ASR",
      "GGUF",
      "Stream"
    ],
    "docs": [
      "docs/asr.md",
      "docs/gguf.md"
    ]
  },
  "package_defaults": {
    "download": {
      "kind": "huggingface_snapshot",
      "repo": "audio-cpp/audio.cpp-gguf",
      "revision": "main",
      "gated": false
    }
  },
  "packages": [
    {
      "id": "higgs_audio_stt_q8_0",
      "display_name": "Higgs Audio v3 STT Q8_0 GGUF",
      "default": true,
      "format": "gguf",
      "precision": "q8_0",
      "target_directory": "Higgs-Audio-v3-STT-GGUF",
      "files": [
        "Higgs-Audio-v3-STT-GGUF/higgs-audio-v3-stt-q8_0.gguf"
      ],
      "strip_prefix": "Higgs-Audio-v3-STT-GGUF"
    },
    {
      "id": "higgs_audio_stt_f16",
      "display_name": "Higgs Audio v3 STT F16 GGUF",
      "format": "gguf",
      "precision": "f16",
      "target_directory": "Higgs-Audio-v3-STT-GGUF",
      "files": [
        "Higgs-Audio-v3-STT-GGUF/higgs-audio-v3-stt-f16.gguf"
      ],
      "strip_prefix": "Higgs-Audio-v3-STT-GGUF"
    }
  ],
  "sources": [
    {
      "format": "gguf",
      "roots": {
        "model": ".",
        "weights": "$gguf"
      },
      "files": {
        "config": "model:config.json",
        "generation_config": "model:generation_config.json",
        "preprocessor_config": "model:preprocessor_config.json",
        "tokenizer_config": "model:tokenizer_config.json",
        "vocab": "model:vocab.json",
        "merges": "model:merges.txt"
      },
      "tensors": {
        "weights": "weights:"
      }
    },
    {
      "format": "safetensors",
      "roots": {
        "model": ".",
        "whisper": "../whisper-large-v3"
      },
      "files": {
        "config": "model:config.json",
        "generation_config": "model:generation_config.json",
        "preprocessor_config": "whisper:preprocessor_config.json",
        "tokenizer_config": "model:tokenizer_config.json",
        "vocab": "model:vocab.json",
        "merges": "model:merges.txt"
      },
      "tensors": {
        "weights": "model:model.safetensors.index.json"
      }
    }
  ]
}