{
  "family": "moss_tts_nano",
  "display_name": "MOSS-TTS-Nano",
  "description": "Compact deployment-first MOSS-TTS model for real-time multilingual speech generation, lightweight integration, and zero-shot voice cloning.",
  "category": "tts",
  "status": "supported",
  "tasks": [
    "tts",
    "clone"
  ],
  "modes": [
    "offline"
  ],
  "languages": [
    "ar",
    "cs",
    "da",
    "de",
    "el",
    "en",
    "es",
    "fa",
    "fr",
    "hu",
    "it",
    "ja",
    "ko",
    "pl",
    "pt",
    "ru",
    "sv",
    "tr",
    "zh"
  ],
  "capabilities": {
    "clone": [
      "speaker_reference"
    ]
  },
  "options": {
    "request": [
      {
        "name": "max_tokens",
        "type": "int",
        "description": "Maximum generated audio frames; default 300.",
        "required": false,
        "min": 0,
        "default": 300
      },
      {
        "name": "do_sample",
        "type": "bool",
        "description": "Enable stochastic audio-token sampling; default true.",
        "required": false,
        "default": true
      },
      {
        "name": "temperature",
        "type": "float",
        "description": "Audio-token sampling temperature; default 1.7.",
        "required": false,
        "min": 0.0,
        "default": 1.7
      },
      {
        "name": "top_p",
        "type": "float",
        "description": "Audio-token nucleus sampling limit; default 0.8.",
        "required": false,
        "min": 0.0,
        "max": 1.0,
        "default": 0.8
      },
      {
        "name": "top_k",
        "type": "int",
        "description": "Audio-token top-k sampling limit; default 25.",
        "required": false,
        "min": 0,
        "default": 25
      },
      {
      "name": "repetition_penalty",
      "type": "float",
      "description": "Audio-token repetition penalty; default 1.0.",
      "required": false,
      "min": 0.0,
      "default": 1.0
    },
    {
      "name": "seed",
      "type": "int",
      "description": "Sampling seed; omitted value uses a random seed.",
      "required": false,
      "min": 0
    },
    {
      "name": "active_codebooks",
      "type": "int",
      "description": "Number of audio tokenizer RVQ codebooks to generate/decode; default is the model-configured codebook count, normally 16.",
      "required": false,
      "min": 1,
      "default": 16
    },
    {
      "name": "reference_text",
      "type": "string",
      "description": "Transcript for reference audio used in voice cloning; requires speaker reference audio.",
      "required": false
    },
    {
      "name": "text_chunk_size",
      "type": "int",
      "description": "Long-form text chunk size in Unicode codepoints; default 256.",
      "required": false,
      "min": 1,
      "default": 256
    },
    {
      "name": "text_temperature",
        "type": "float",
        "description": "Text-gate sampling temperature; default 1.5.",
        "required": false,
        "default": 1.5
      },
      {
        "name": "text_top_p",
        "type": "float",
        "description": "Text-gate nucleus sampling limit; default 1.0.",
        "required": false,
        "default": 1.0
      },
      {
        "name": "text_top_k",
        "type": "int",
        "description": "Text-gate top-k sampling limit; default 50.",
        "required": false,
        "default": 50
      },
      {
      "name": "text_chunk_mode",
      "type": "enum",
      "description": "Long-form text chunking mode; default word_budget. Text chunk size defaults to 256 Unicode codepoints.",
      "preset": "text_chunk_mode_full",
      "required": false,
        "default": "word_budget"
      }
    ],
    "session": [
      {
      "name": "weight_type",
      "type": "enum",
      "description": "Global and local-frame weight storage type; default native.",
      "preset": "weight_type_full",
      "required": false,
      "default": "native"
    },
    {
      "name": "global_weight_type",
      "type": "enum",
      "description": "Global transformer weight storage type; defaults to weight_type, which defaults to native.",
      "preset": "weight_type_full",
      "required": false
    },
    {
      "name": "local_frame_weight_type",
      "type": "enum",
      "description": "Local frame decoder weight storage type; defaults to weight_type, which defaults to native.",
      "preset": "weight_type_full",
      "required": false
      },
      {
        "name": "global_prefill_graph_arena_mb",
        "type": "int",
        "description": "Global transformer prefill graph arena size in MiB; default 256.",
        "required": false,
        "min": 0,
        "default": 256
      },
      {
        "name": "global_decode_graph_arena_mb",
        "type": "int",
        "description": "Global transformer decode graph arena size in MiB; default 128.",
        "required": false,
        "min": 0,
        "default": 128
      },
      {
        "name": "global_weight_context_mb",
        "type": "int",
        "description": "Global transformer weight context size in MiB; default 512.",
        "required": false,
        "min": 0,
        "default": 512
      },
      {
        "name": "local_frame_graph_arena_mb",
        "type": "int",
        "description": "Local frame decoder graph arena size in MiB; default 64.",
        "required": false,
        "min": 0,
        "default": 64
      },
      {
        "name": "local_frame_weight_context_mb",
        "type": "int",
        "description": "Local frame decoder weight context size in MiB; default 128.",
        "required": false,
        "min": 0,
        "default": 128
      },
      {
        "name": "audio_tokenizer_encoder_graph_arena_mb",
        "type": "int",
        "description": "Audio tokenizer encoder graph arena size in MiB; default 64.",
        "required": false,
        "min": 0,
        "default": 64
      },
      {
        "name": "audio_tokenizer_decoder_graph_arena_mb",
        "type": "int",
        "description": "Audio tokenizer decoder graph arena size in MiB; default 64.",
        "required": false,
        "min": 0,
        "default": 64
      },
      {
        "name": "audio_tokenizer_weight_context_mb",
        "type": "int",
        "description": "Audio tokenizer weight context size in MiB; default 128.",
        "required": false,
        "min": 0,
        "default": 128
      }
    ],
    "load": []
  },
  "runtime": {
    "tags": [
      "gguf"
    ]
  },
  "sources": [
    {
      "format": "gguf",
      "roots": {
        "model": ".",
        "weights": "$gguf"
      },
      "files": {
        "config": "model:config.json",
        "tokenizer_config": "model:tokenizer_config.json",
        "tokenizer_model": "model:tokenizer.model",
        "audio_tokenizer_config": "model:audio_tokenizer/config.json"
      },
      "tensors": {
        "model_weights": {
          "source": "weights:",
          "prefix": "model_weights"
        },
        "audio_tokenizer_weights": {
          "source": "weights:",
          "prefix": "audio_tokenizer_weights"
        }
      }
    },
    {
      "format": "safetensors",
      "roots": {
        "model": ".",
        "audio_tokenizer": "audio_tokenizer"
      },
      "files": {
        "config": "model:config.json",
        "tokenizer_config": "model:tokenizer_config.json",
        "tokenizer_model": "model:tokenizer.model",
        "audio_tokenizer_config": "audio_tokenizer:config.json"
      },
      "tensors": {
        "model_weights": "model:model.safetensors",
        "audio_tokenizer_weights": "audio_tokenizer:model.safetensors.index.json"
      }
    }
  ],
  "package_defaults": {
    "download": {
      "kind": "huggingface_snapshot",
      "repo": "audio-cpp/audio.cpp-gguf",
      "revision": "main",
      "gated": false
    }
  },
  "packages": [
    {
      "id": "moss_tts_nano_100m_q8_0",
      "display_name": "MOSS-TTS-Nano 100M Q8_0 GGUF",
      "default": true,
      "format": "gguf",
      "precision": "q8_0",
      "target_directory": "MOSS-TTS-Nano-100M-GGUF",
      "files": [
        "MOSS-TTS-Nano-100M-GGUF/moss-tts-nano-100m-q8_0.gguf"
      ],
      "strip_prefix": "MOSS-TTS-Nano-100M-GGUF"
    },
    {
      "id": "moss_tts_nano_100m_bf16",
      "display_name": "MOSS-TTS-Nano 100M BF16 GGUF",
      "format": "gguf",
      "precision": "bf16",
      "target_directory": "MOSS-TTS-Nano-100M-GGUF",
      "files": [
        "MOSS-TTS-Nano-100M-GGUF/moss-tts-nano-100m-bf16.gguf"
      ],
      "strip_prefix": "MOSS-TTS-Nano-100M-GGUF"
    }
  ]
}