{
  "model_count": 5,
  "models": [
    {
      "audio": {
        "formats": [
          "wav",
          "pcm"
        ],
        "sample_rate_hz": 24000
      },
      "available": false,
      "capabilities": {
        "formats": [
          "wav",
          "pcm"
        ],
        "long_form": "limited",
        "multilingual": "full",
        "streaming": false,
        "style_control": [
          "exaggeration",
          "cfg_weight",
          "paralinguistic_tags"
        ],
        "voice_cloning": true
      },
      "executable": false,
      "id": "chatterbox:tts",
      "manifest_path": "manifests/chatterbox.yaml",
      "name": "Chatterbox TTS",
      "notes": [
        "Manifest-only example for expressive and cloning-oriented TTS.",
        "Chatterbox outputs include Resemble AI PerTh watermarking in official examples."
      ],
      "policy": {
        "commercial_use": true,
        "requires_voice_consent": true
      },
      "requires_extra": null,
      "runtime": {
        "acceleration": {
          "cpu": false,
          "cuda": true,
          "metal": false
        },
        "kind": "python",
        "package": "chatterbox-tts"
      },
      "schema_version": "0.1",
      "status": "manifest-only",
      "upstream": {
        "homepage": "https://github.com/resemble-ai/chatterbox",
        "license": "mit",
        "weights": "https://huggingface.co/ResembleAI/chatterbox"
      },
      "voices": {
        "builtin": false,
        "custom": true
      }
    },
    {
      "audio": {
        "formats": [
          "mp3",
          "wav",
          "pcm"
        ],
        "sample_rate_hz": 24000
      },
      "available": true,
      "capabilities": {
        "formats": [
          "mp3",
          "wav",
          "pcm"
        ],
        "long_form": "limited",
        "multilingual": "limited",
        "streaming": false,
        "style_control": "speed",
        "voice_cloning": false
      },
      "executable": true,
      "id": "fake:tts",
      "manifest_path": "manifests/fake-tts.yaml",
      "name": "TimbreGrid Fake TTS",
      "notes": [
        "Deterministic test adapter for conformance and benchmark development."
      ],
      "policy": {
        "commercial_use": true,
        "requires_voice_consent": false
      },
      "requires_extra": null,
      "runtime": {
        "acceleration": {
          "cpu": true,
          "cuda": false,
          "metal": false
        },
        "kind": "python",
        "package": "timbregrid"
      },
      "schema_version": "0.1",
      "status": "available",
      "upstream": {
        "homepage": "https://github.com/kiyeonjeon21/timbregrid",
        "license": "mit",
        "weights": "none"
      },
      "voices": {
        "builtin": true,
        "custom": false
      }
    },
    {
      "audio": {
        "formats": [
          "wav",
          "pcm"
        ],
        "sample_rate_hz": 24000
      },
      "available": false,
      "capabilities": {
        "formats": [
          "wav",
          "pcm"
        ],
        "long_form": "limited",
        "multilingual": "none",
        "streaming": false,
        "style_control": "speed",
        "voice_cloning": false
      },
      "executable": true,
      "id": "kitten-tts:nano-0.8",
      "manifest_path": "manifests/kitten-tts-nano-0.8.yaml",
      "name": "KittenTTS Nano 0.8",
      "notes": [
        "Optional executable adapter for lightweight ONNX/CPU edge TTS.",
        "Official docs describe 15M to 80M parameter variants and built-in voices."
      ],
      "policy": {
        "commercial_use": true,
        "requires_voice_consent": false
      },
      "requires_extra": "kittentts",
      "runtime": {
        "acceleration": {
          "cpu": true,
          "cuda": false,
          "metal": false
        },
        "kind": "python",
        "package": "kittentts"
      },
      "schema_version": "0.1",
      "status": "requires optional dependency: kittentts",
      "upstream": {
        "homepage": "https://github.com/KittenML/KittenTTS",
        "license": "apache-2.0",
        "weights": "https://huggingface.co/KittenML/kitten-tts-nano-0.8-fp32"
      },
      "voices": {
        "builtin": true,
        "custom": false
      }
    },
    {
      "audio": {
        "formats": [
          "wav",
          "pcm"
        ],
        "sample_rate_hz": 24000
      },
      "available": false,
      "capabilities": {
        "formats": [
          "wav",
          "pcm"
        ],
        "long_form": "limited",
        "multilingual": "limited",
        "streaming": false,
        "style_control": "speed",
        "voice_cloning": false
      },
      "executable": true,
      "id": "kokoro:82m",
      "manifest_path": "manifests/kokoro-82m.yaml",
      "name": "Kokoro 82M",
      "notes": [
        "Good baseline for low-latency local TTS.",
        "Executable through the optional timbregrid[kokoro] adapter."
      ],
      "policy": {
        "commercial_use": true,
        "requires_voice_consent": false
      },
      "requires_extra": "kokoro",
      "runtime": {
        "acceleration": {
          "cpu": true,
          "cuda": true,
          "metal": "optional"
        },
        "kind": "python",
        "package": "kokoro"
      },
      "schema_version": "0.1",
      "status": "requires optional dependency: kokoro",
      "upstream": {
        "homepage": "https://huggingface.co/hexgrad/Kokoro-82M",
        "license": "apache-2.0",
        "weights": "open-weight"
      },
      "voices": {
        "builtin": true,
        "custom": false
      }
    },
    {
      "audio": {
        "formats": [
          "wav",
          "pcm"
        ],
        "sample_rate_hz": 24000
      },
      "available": false,
      "capabilities": {
        "formats": [
          "wav",
          "pcm"
        ],
        "long_form": "limited",
        "multilingual": "full",
        "streaming": true,
        "style_control": [
          "voice_clone",
          "voice_description"
        ],
        "voice_cloning": true
      },
      "executable": false,
      "id": "qwen3-tts:0.6b-base",
      "manifest_path": "manifests/qwen3-tts-0.6b-base.yaml",
      "name": "Qwen3-TTS 12Hz 0.6B Base",
      "notes": [
        "Manifest-only example for multilingual, streaming, and voice-cloning TTS.",
        "Base checkpoint supports rapid cloning from user-provided reference audio."
      ],
      "policy": {
        "commercial_use": true,
        "requires_voice_consent": true
      },
      "requires_extra": null,
      "runtime": {
        "acceleration": {
          "cpu": false,
          "cuda": true,
          "metal": false
        },
        "kind": "python",
        "package": "qwen-tts"
      },
      "schema_version": "0.1",
      "status": "manifest-only",
      "upstream": {
        "homepage": "https://huggingface.co/Qwen/Qwen3-TTS-12Hz-0.6B-Base",
        "license": "apache-2.0",
        "weights": "https://huggingface.co/Qwen/Qwen3-TTS-12Hz-0.6B-Base"
      },
      "voices": {
        "builtin": false,
        "custom": true
      }
    }
  ],
  "schema_version": "0.1"
}