{
  "hf_id": "OpenMOSS-Team/MOSS-SoundEffect",
  "meta": {
    "title": "MOSS-SoundEffect",
    "slug": "moss-soundeffect",
    "provider": "OpenMOSS",
    "description": "OpenMOSS's 8B sound-effect generation model — environmental, urban, biological, human-action and musical sounds with controllable duration, no reference audio — served via vLLM-Omni through the OpenAI /v1/audio/speech API (24 kHz mono).",
    "date_added": "2026-06-17",
    "date_updated": "2026-06-15",
    "tasks": [
      "omni"
    ],
    "related_recipes": [
      "OpenMOSS-Team/MOSS-TTS",
      "stabilityai/stable-audio-open-1.0",
      "OpenMOSS-Team/MOSS-VoiceGenerator"
    ]
  },
  "model": {
    "model_id": "OpenMOSS-Team/MOSS-SoundEffect",
    "min_vllm_version": "0.22.0",
    "architecture": "dense",
    "parameter_count": "8B",
    "active_parameters": "8B",
    "context_length": 0,
    "base_args": [
      "--omni",
      "--deploy-config",
      "vllm_omni/deploy/moss_sound_effect.yaml"
    ],
    "base_env": {},
    "install": {
      "pip": {
        "command": "uv venv\nsource .venv/bin/activate\nuv pip install -U vllm --torch-backend auto"
      },
      "docker": {
        "command": "docker pull vllm/vllm-openai:latest"
      },
      "extras": [
        {
          "note": "vLLM-Omni provides the audio serving backend and the `--omni` serve path. See the installation guide for pre-built wheels or building from source.",
          "command": "uv pip install git+https://github.com/vllm-project/vllm-omni.git"
        }
      ]
    }
  },
  "omni": {
    "tasks": [
      {
        "id": "t2a",
        "label": "Text → Sound",
        "endpoint": "/v1/audio/speech",
        "description": "MOSS-SoundEffect sound-effect synthesis from a text description (no ref audio)",
        "curl": "curl -X POST http://localhost:8000/v1/audio/speech \\\n  -H \"Content-Type: application/json\" \\\n  -d '{\n    \"model\": \"OpenMOSS-Team/MOSS-SoundEffect\",\n    \"input\": \"Thunder rumbling, rain pattering on a tin roof.\",\n    \"response_format\": \"wav\"\n  }' --output thunder.wav\n"
      }
    ]
  },
  "features": {},
  "opt_in_features": [],
  "variants": {
    "default": {
      "precision": "bf16",
      "vram_minimum_gb": 32,
      "description": "MOSS-SoundEffect 8B (MossTTSDelay, n_vq=16) — text-to-sound-effect"
    }
  },
  "hardware_overrides": {},
  "guide": "## Overview\n\n[MOSS-SoundEffect](https://huggingface.co/OpenMOSS-Team/MOSS-SoundEffect) is the\n**sound-effect generation** member of OpenMOSS's MOSS-TTS Family, served through\n**vLLM-Omni** with the OpenAI-compatible `/v1/audio/speech` API. It generates\naudio for natural environments, urban scenes, biological sounds, human actions,\nand musical fragments — suitable for film, games, and interactive experiences.\nOutput is **24 kHz mono**.\n\nUnlike the speech members of the family, MOSS-SoundEffect takes **no reference\naudio**: the `input` field is a text description of the sound (mapped to the\nupstream `ambient_sound` parameter). Generation runs at ~12.5 tokens/second;\nlonger descriptions (or a higher token budget) produce longer audio.\n\n## Prerequisites\n\n- **Hardware**: a single CUDA GPU comparable to the 8B MOSS-TTS profile\n  (~18 GB talker + ~8 GB codec).\n- **vLLM-Omni** targeting vLLM >= 0.22.\n\n## Installation\n\n```bash\nuv venv && source .venv/bin/activate\nuv pip install git+https://github.com/vllm-project/vllm-omni.git\n```\n\n## Launch the server\n\nMOSS-SoundEffect shares `model_type=moss_tts_delay`, so pass its deploy config\nexplicitly:\n\n```bash\nvllm serve OpenMOSS-Team/MOSS-SoundEffect --omni \\\n  --deploy-config vllm_omni/deploy/moss_sound_effect.yaml --port 8000\n```\n\n## Client usage\n\n```bash\ncurl -X POST http://localhost:8000/v1/audio/speech \\\n  -H \"Content-Type: application/json\" \\\n  -d '{\n    \"model\": \"OpenMOSS-Team/MOSS-SoundEffect\",\n    \"input\": \"Thunder rumbling, rain pattering on a tin roof.\",\n    \"response_format\": \"wav\"\n  }' --output thunder.wav\n```\n\n## Known limitations\n\n- Output is **24 kHz mono**.\n- No `ref_audio` is accepted; `input` is a text description of the sound.\n\n## References\n\n- [Model card](https://huggingface.co/OpenMOSS-Team/MOSS-SoundEffect)\n- [MOSS-TTS Family](https://huggingface.co/OpenMOSS-Team/MOSS-TTS) · [GitHub](https://github.com/OpenMOSS/MOSS-TTS)\n",
  "source": {
    "source": "vllm-omni",
    "repo": "vllm-project/vllm-omni",
    "ref": "main",
    "sha": "beeedd99df33b432b7cf74caa8b28e1893124827",
    "path": "recipes/models/OpenMOSS-Team/MOSS-SoundEffect.yaml",
    "edit_url": "https://github.com/vllm-project/vllm-omni/edit/main/recipes/models/OpenMOSS-Team/MOSS-SoundEffect.yaml",
    "overlays": []
  }
}