voice_audio_multi/model_specs/canary_asr.json

{
  "family": "canary_asr",
  "schema_version": 1,
  "display_name": "Canary 180M Flash",
  "description": "NVIDIA Canary multilingual speech recognition and speech translation.",
  "category": "asr",
  "status": "supported",
  "tasks": [
    "asr"
  ],
  "modes": [
    "offline"
  ],
  "languages": [
    "en",
    "de",
    "es",
    "fr"
  ],
  "capabilities": {},
  "dependencies": [],
  "options": {
    "request": [
      {
        "name": "language",
        "type": "string",
        "description": "Input audio language.",
        "required": false,
        "default": "en"
      },
      {
        "name": "target_language",
        "type": "string",
        "description": "Output language: en, de, es, or fr. Defaults to the input language for transcription.",
        "required": false
      },
      {
        "name": "punc",
        "type": "bool",
        "description": "Generate punctuation and capitalization.",
        "required": false,
        "default": true
      },
      {
        "name": "max_tokens",
        "type": "int",
        "description": "Maximum generated tokens per chunk; zero uses the reference encoder-length-based limit.",
        "required": false,
        "min": 0,
        "max": 1015,
        "default": 0
      },
      {
        "name": "audio_chunk_mode",
        "type": "enum",
        "description": "Long-form audio chunking mode.",
        "values": [
          "auto",
          "fixed",
          "none"
        ],
        "required": false,
        "default": "auto"
      },
      {
        "name": "audio_chunk_duration_sec",
        "type": "float",
        "description": "Maximum duration of each audio chunk in seconds.",
        "required": false,
        "min": 0.02,
        "max": 40,
        "default": 40
      }
    ],
    "session": [
      {
        "name": "weight_type",
        "type": "enum",
        "description": "Matmul weight storage type.",
        "values": [
          "native",
          "f32",
          "f16",
          "bf16",
          "q8_0",
          "q4_0",
          "q4_1",
          "q5_0",
          "q5_1",
          "q2_k",
          "q3_k",
          "q4_k",
          "q5_k",
          "q6_k"
        ],
        "required": false,
        "default": "native"
      }
    ],
    "load": []
  },
  "runtime": {
    "tags": [
      "gguf",
      "cpu",
      "cuda"
    ]
  },
  "ui": {
    "tags": [
      "ASR",
      "GGUF"
    ],
    "docs": [
      "docs/models/canary_asr.md"
    ],
    "recommended_package": "canary_180m_flash_f32"
  },
  "package_defaults": {
    "download": {
      "kind": "huggingface_snapshot",
      "repo": "audio-cpp/audio.cpp-gguf",
      "revision": "main",
      "gated": false
    }
  },
  "packages": [
    {
      "id": "canary_180m_flash_f32",
      "display_name": "Canary 180M Flash GGUF F32",
      "default": true,
      "format": "gguf",
      "precision": "f32",
      "target_directory": "Canary-180M-Flash-GGUF",
      "files": [
        "Canary-180M-Flash-GGUF/canary-180m-flash-f32.gguf"
      ],
      "strip_prefix": "Canary-180M-Flash-GGUF"
    },
    {
      "id": "canary_180m_flash_q8_0",
      "display_name": "Canary 180M Flash GGUF Q8_0",
      "format": "gguf",
      "precision": "q8_0",
      "target_directory": "Canary-180M-Flash-GGUF",
      "files": [
        "Canary-180M-Flash-GGUF/canary-180m-flash-q8_0.gguf"
      ],
      "strip_prefix": "Canary-180M-Flash-GGUF"
    }
  ],
  "sources": [
    {
      "format": "safetensors",
      "roots": {
        "model": "."
      },
      "files": {
        "config": "model:config.json",
        "tokenizer_spl_tokens": "model:tokenizer_spl_tokens.model",
        "tokenizer_en": "model:tokenizer_en.model",
        "tokenizer_de": "model:tokenizer_de.model",
        "tokenizer_es": "model:tokenizer_es.model",
        "tokenizer_fr": "model:tokenizer_fr.model"
      },
      "tensors": {
        "weights": "model:model.safetensors"
      }
    },
    {
      "format": "gguf",
      "roots": {
        "model": ".",
        "weights": "$gguf"
      },
      "files": {
        "config": "model:config.json",
        "tokenizer_spl_tokens": "model:tokenizer_spl_tokens.model",
        "tokenizer_en": "model:tokenizer_en.model",
        "tokenizer_de": "model:tokenizer_de.model",
        "tokenizer_es": "model:tokenizer_es.model",
        "tokenizer_fr": "model:tokenizer_fr.model"
      },
      "tensors": {
        "weights": "weights:"
      }
    }
  ]
}