voice_audio_multi/model_specs/midashenglm_gen.json

{
  "schema_version": 1,
  "family": "midashenglm_gen",
  "display_name": "MiDashengLM-Gen",
  "description": "DashengLM generative audio model for speech, music, sound effects, and ambience prompts.",
  "category": "audio_generation",
  "status": "experimental",
  "tasks": [
    "music"
  ],
  "modes": [
    "offline"
  ],
  "languages": [
    "auto"
  ],
  "runtime": {
    "tags": [
      "gguf",
      "cuda"
    ]
  },
  "capabilities": {
    "music": [
      "instrumental",
      "style_control"
    ]
  },
  "options": {
    "request": [
      {
        "name": "duration_sec",
        "type": "float",
        "description": "Target audio duration budget; omitted derives from the checkpoint sequence length, hop size and sample rate. Must be finite and positive; no fixed runtime upper cap.",
        "required": false,
        "default": 20
      },
      {
        "name": "guidance_scale",
        "type": "float",
        "description": "Classifier-free guidance scale. Any finite value is accepted.",
        "required": false,
        "default": 2
      },
      {
        "name": "stop_threshold",
        "type": "float",
        "description": "Stop probability threshold. Any finite value is accepted.",
        "required": false,
        "default": 0.5
      },
      {
        "name": "min_stop_step",
        "type": "int",
        "description": "Minimum generated steps before accepting stop.",
        "required": false,
        "min": 1,
        "default": 5
      },
      {
        "name": "seed",
        "type": "int",
        "description": "Unsigned 32-bit generation seed; default 0 is a fixed seed. Negative values are not accepted.",
        "required": false,
        "min": 0,
        "max": 4294967295,
        "default": 0
      }
    ],
    "session": [
      {
        "name": "weight_type",
        "type": "enum",
        "description": "Shared weight storage type.",
        "preset": "weight_type_full",
        "required": false,
        "default": "native"
      }
    ],
    "load": []
  },
  "package_defaults": {
    "download": {
      "kind": "huggingface_snapshot",
      "repo": "audio-cpp/audio.cpp-gguf",
      "revision": "main",
      "gated": false
    }
  },
  "packages": [
    {
      "id": "midashenglm_gen_f32",
      "display_name": "MiDashengLM-Gen F32 GGUF",
      "format": "gguf",
      "precision": "f32",
      "target_directory": "MiDashengLM-Gen-GGUF",
      "files": [
        "MiDashengLM-Gen-GGUF/midashenglm-gen-f32.gguf"
      ],
      "strip_prefix": "MiDashengLM-Gen-GGUF"
    },
    {
      "id": "midashenglm_gen_q8_0",
      "display_name": "MiDashengLM-Gen Q8_0 GGUF",
      "default": true,
      "format": "gguf",
      "precision": "q8_0",
      "target_directory": "MiDashengLM-Gen-GGUF",
      "files": [
        "MiDashengLM-Gen-GGUF/midashenglm-gen-q8_0.gguf"
      ],
      "strip_prefix": "MiDashengLM-Gen-GGUF"
    }
  ],
  "dependencies": [],
  "ui": {
    "recommended_package": "midashenglm_gen_q8_0",
    "tags": [
      "Music",
      "SFX",
      "GGUF"
    ],
    "docs": [
      "docs/models/midashenglm_gen.md",
      "docs/gguf.md"
    ]
  },
  "sources": [
    {
      "format": "gguf",
      "roots": {
        "model": ".",
        "weights": "$gguf"
      },
      "files": {
        "config": "model:config.json",
        "vocab.json": "model:vocab.json",
        "merges.txt": "model:merges.txt",
        "tokenizer_config.json": "model:tokenizer_config.json"
      },
      "tensors": {
        "weights": "weights:"
      }
    },
    {
      "format": "safetensors",
      "roots": {
        "model": "."
      },
      "files": {
        "config": "model:config.json",
        "vocab.json": "model:vocab.json",
        "merges.txt": "model:merges.txt",
        "tokenizer_config.json": "model:tokenizer_config.json"
      },
      "tensors": {
        "weights": "model:model.safetensors.index.json"
      }
    }
  ]
}