jamiepine/voicebox · error · ValueError

Unknown model size: {model_size}

Error message

Unknown model size: {model_size}

What it means

Raised by PyTorchTTSBackend._get_model_path when model_size is not in hf_model_map, whose only keys are "1.7B" (Qwen/Qwen3-TTS-12Hz-1.7B-Base) and "0.6B" (Qwen/Qwen3-TTS-12Hz-0.6B-Base). The map maps a size token to the HuggingFace Hub repo for the PyTorch Qwen3-TTS base weights.

Source

Thrown at backend/backends/pytorch_backend.py:59

        return self.model is not None

    def _get_model_path(self, model_size: str) -> str:
        """
        Get the HuggingFace Hub model ID.

        Args:
            model_size: Model size (1.7B or 0.6B)

        Returns:
            HuggingFace Hub model ID
        """
        hf_model_map = {
            "1.7B": "Qwen/Qwen3-TTS-12Hz-1.7B-Base",
            "0.6B": "Qwen/Qwen3-TTS-12Hz-0.6B-Base",
        }

        if model_size not in hf_model_map:
            raise ValueError(f"Unknown model size: {model_size}")

        return hf_model_map[model_size]

    def _is_model_cached(self, model_size: str) -> bool:
        return is_model_cached(self._get_model_path(model_size))

    async def load_model_async(self, model_size: Optional[str] = None):
        """
        Lazy load the TTS model with automatic downloading from HuggingFace Hub.

        Args:
            model_size: Model size to load (1.7B or 0.6B)
        """
        if model_size is None:
            model_size = self.model_size

        # If already loaded with correct size, return
        if self.model is not None and self._current_model_size == model_size:

View on GitHub (pinned to 51f49dea19)

Solutions

1. Use "1.7B" or "0.6B" exactly.
2. Whitelist/normalize at the API boundary so the backend only ever sees valid tokens.
3. Keep TTS size constants separate from LLM size constants to avoid "4B" leaking through.

Example fix

// before
backend.load_model_async(model_size="4B")
// after
backend.load_model_async(model_size="0.6B")

PYTORCH_TTS_SIZES = {"1.7B", "0.6B"}
if model_size not in PYTORCH_TTS_SIZES:
    raise ValueError(f"model_size must be one of {sorted(PYTORCH_TTS_SIZES)}")
await backend.load_model_async(model_size=model_size)

def is_pytorch_tts_size(value: str) -> bool:
    return isinstance(value, str) and value in {"1.7B", "0.6B"}

try:
    await backend.load_model_async(model_size=model_size)
except ValueError as exc:
    if "Unknown model size" in str(exc):
        model_size = "1.7B"
        await backend.load_model_async(model_size=model_size)
    else:
        raise

When it happens

Trigger: Calling load_model_async or _get_model_path with a value other than "1.7B" or "0.6B" — e.g. "4B", lowercase "1.7b", or "large".

Common situations: Sharing one model-size constant across engines where only some accept it; passing the LLM backend's "4B" to the TTS backend; untrimmed UI input; mistaking whisper-style names ("base"/"turbo") for TTS sizes.

Related errors

• Unknown model size: {model_size}
• Unknown model size: {model_size}
• Unknown Qwen3 size: {model_size}
• Invalid STT model '{model_size}'. Must be one of: {', '.join
• RPC ${method}: ${json.error.message}

AI-assisted analysis of jamiepine/voicebox@51f49dea19 (2026-08-12). Data as JSON: /api/errors/511c6d117a02d6f5. Report an issue: GitHub.