{
  "family": "higgs_audio_tts",
  "display_name": "Higgs Audio v3 TTS",
  "description": "Boson AI conversational TTS model for expressive speech across 100+ languages, zero-shot voice cloning, and inline control over emotion, style, prosody, pauses, and sound effects.",
  "category": "tts",
  "status": "supported",
  "tasks": [
    "tts",
    "clone"
  ],
  "modes": [
    "offline"
  ],
  "languages": [
    "100+ languages"
  ],
  "capabilities": {
    "clone": [
      "speaker_reference"
    ]
  },
  "options": {
    "request": [
      {
        "name": "max_tokens",
        "type": "int",
        "description": "Maximum generated AR tokens; default 2048, and 0 also uses the default.",
        "required": false,
        "min": 0,
        "default": 2048
      },
      {
        "name": "temperature",
        "type": "float",
        "description": "AR sampling temperature; must be positive when set; default 0.8.",
        "required": false,
        "min": 0.0,
        "default": 0.8
      },
      {
        "name": "top_k",
        "type": "int",
        "description": "AR top-k sampling limit; default 30.",
        "required": false,
        "min": 0,
        "default": 30
      },
      {
        "name": "top_p",
        "type": "float",
        "description": "AR nucleus sampling probability in (0, 1]; default 0.8.",
        "required": false,
        "min": 0.0,
        "max": 1.0,
        "default": 0.8
      },
      {
        "name": "repetition_penalty",
        "type": "float",
        "description": "Accepted for API compatibility and validated as positive; current Higgs audio sampling does not consume it; default 1.1.",
        "required": false,
        "min": 0.0,
        "default": 1.1
      },
      {
        "name": "seed",
        "type": "int",
        "description": "Generation seed for reproducible output; omitted uses a random seed.",
        "required": false,
        "min": 0
      },
      {
        "name": "reference_text",
        "type": "string",
        "description": "Transcript or prompt text paired with reference audio for zero-shot voice cloning.",
        "required": false
      },
      {
        "name": "text_chunk_size",
        "type": "int",
        "description": "Maximum characters per generated text chunk; default 1024.",
        "required": false,
        "min": 1,
        "default": 1024
      },
      {
        "name": "text_chunk_mode",
        "type": "enum",
        "description": "Framework text chunking mode; default word_budget.",
        "preset": "text_chunk_mode_full",
        "required": false,
        "default": "word_budget"
      }
    ],
    "session": [
      {
        "name": "weight_type",
        "type": "enum",
        "description": "AR and codec weight storage type; default native.",
        "preset": "weight_type_full",
        "required": false,
        "default": "native"
      },
      {
        "name": "ar_weight_type",
        "type": "enum",
        "description": "Autoregressive decoder weight storage type; defaults to weight_type when set, otherwise native.",
        "preset": "weight_type_full",
        "required": false
      },
      {
        "name": "codec_weight_type",
        "type": "enum",
        "description": "Codec weight storage type; defaults to weight_type when set, otherwise native.",
        "preset": "weight_type_full",
        "required": false
      },
      {
        "name": "ar_weight_context_mb",
        "type": "int",
        "description": "AR weight context size in MiB; default 4096.",
        "required": false,
        "min": 0,
        "default": 4096
      },
      {
        "name": "codec_weight_context_mb",
        "type": "int",
        "description": "Codec weight context size in MiB; default 1536.",
        "required": false,
        "min": 0,
        "default": 1536
      },
      {
        "name": "ar_decode_graph_arena_mb",
        "type": "int",
        "description": "AR decode graph arena size in MiB; default 512.",
        "required": false,
        "min": 0,
        "default": 512
      },
      {
        "name": "codec_decode_graph_arena_mb",
        "type": "int",
        "description": "Codec decode graph arena size in MiB; default 128.",
        "required": false,
        "min": 0,
        "default": 128
      },
      {
        "name": "codec_encode_graph_arena_mb",
        "type": "int",
        "description": "Codec encode graph arena size in MiB; default 256.",
        "required": false,
        "min": 0,
        "default": 256
      },
      {
        "name": "reference_cache_slots",
        "type": "int",
        "description": "Encoded reference-audio cache slots; default 1.",
        "required": false,
        "min": 0,
        "default": 1
      }
    ],
    "load": []
  },
  "runtime": {
    "tags": [
      "gguf"
    ]
  },
  "sources": [
    {
      "format": "gguf",
      "roots": {
        "model": ".",
        "weights": "$gguf"
      },
      "files": {
        "config": "model:config.json",
        "tokenizer_json": "model:tokenizer.json",
        "tokenizer_config": "model:tokenizer_config.json",
        "chat_template": "model:chat_template.jinja"
      },
      "tensors": {
        "weights": "weights:"
      }
    },
    {
      "format": "safetensors",
      "roots": {
        "model": "."
      },
      "files": {
        "config": "model:config.json",
        "tokenizer_json": "model:tokenizer.json",
        "tokenizer_config": "model:tokenizer_config.json",
        "chat_template": "model:chat_template.jinja"
      },
      "tensors": {
        "weights": "model:model.safetensors.index.json"
      }
    }
  ],
  "package_defaults": {
    "download": {
      "kind": "huggingface_snapshot",
      "repo": "audio-cpp/audio.cpp-gguf",
      "revision": "main",
      "gated": false
    }
  },
  "packages": [
    {
      "id": "higgs_audio_tts_4b_q8_0",
      "display_name": "Higgs Audio v3 TTS 4B Q8_0 GGUF",
      "default": true,
      "format": "gguf",
      "precision": "q8_0",
      "target_directory": "Higgs-Audio-v3-TTS-4B-GGUF",
      "files": [
        "Higgs-Audio-v3-TTS-4B-GGUF/higgs-audio-v3-tts-4b-q8_0.gguf"
      ],
      "strip_prefix": "Higgs-Audio-v3-TTS-4B-GGUF"
    },
    {
      "id": "higgs_audio_tts_4b_bf16",
      "display_name": "Higgs Audio v3 TTS 4B BF16 GGUF",
      "format": "gguf",
      "precision": "bf16",
      "target_directory": "Higgs-Audio-v3-TTS-4B-GGUF",
      "files": [
        "Higgs-Audio-v3-TTS-4B-GGUF/higgs-audio-v3-tts-4b-bf16.gguf"
      ],
      "strip_prefix": "Higgs-Audio-v3-TTS-4B-GGUF"
    }
  ]
}