# sherpa-onnx-tts Skill

Provides text-to-speech functionality using the sherpa-onnx offline TTS engine with a German female voice (vits-piper-de_DE-eva_k-x_low).

## Installation

This skill is already installed in the workspace. It uses the sherpa-onnx runtime and model files located at:

- Runtime: `~/.openclaw/tools/sherpa-onnx-tts/runtime`
- Model: `~/.openclaw/tools/sherpa-onnx-tts/models/vits-piper-de_DE-eva_k-x_low`

## Usage

The skill provides a command `sherpa-onnx-tts` that can be used to synthesize speech from text.

### Example

```sh
sherpa-onnx-tts -o output.wav "Hallo Sir, dies ist ein Test."
```

### Environment Variables

The skill respects the following environment variables:

- `SHERPA_ONNX_RUNTIME_DIR`: Path to the sherpa-onnx runtime directory (default: `~/.openclaw/tools/sherpa-onnx-tts/runtime`)
- `SHERPA_ONNX_MODEL_DIR`: Path to the model directory (default: `~/.openclaw/tools/sherpa-onnx-tts/models/vits-piper-de_DE-eva_k-x_low`)

## Notes

- The voice is German female (Eva).
- The generated audio is in WAV format (16kHz, mono, 16-bit PCM).
- For best results, use clear German text.

## Troubleshooting

If you encounter errors about missing tokens or model files, ensure the environment variables are set correctly or that the default paths exist.