second-state/qwen3_tts_rs / skills
audio-tts
VerifiedGenerate speech audio from text using Qwen3 TTS, or clone a voice from reference audio. Triggered when the user wants to convert text to speech, generate audio, read text aloud, or clone/mimic a voice. Supports multiple speakers, English and Chinese, and emotion/style control.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 6.3 KB |
| bootstrap.shexecutable | script | 5.0 KB |
| install.md | markdown | 3.5 KB |
install.md
View on GitHub# Qwen3 TTS Skill — Installation Guide
Follow these steps to install the Qwen3 TTS skill that provides text-to-speech and voice cloning capabilities.
## Prerequisites
- `curl` (for downloading)
- `unzip` (for extraction)
- `bash` (shell)
- `git` (for cloning)
- `pip` with `huggingface_hub` and `transformers` (for model download and tokenizer generation)
## Quick Install (Recommended)
```bash
SKILL_DIR="${HOME}/.openclaw/skills/audio_tts"
mkdir -p "$SKILL_DIR"
# Clone the repo
git clone --depth 1 https://github.com/second-state/qwen3_tts_rs.git /tmp/qwen3-tts-repo
cp -r /tmp/qwen3-tts-repo/skills/* "$SKILL_DIR"
mkdir -p "$SKILL_DIR/scripts"
cp -r /tmp/qwen3-tts-repo/reference_audio "$SKILL_DIR/scripts/reference_audio"
rm -rf /tmp/qwen3-tts-repo
# Download platform-specific binaries, libtorch, and models
"${SKILL_DIR}/bootstrap.sh"
```
After installation, verify it works:
```bash
# Test TTS (requires model download to complete)
~/.openclaw/skills/audio_tts/scripts/tts \
~/.openclaw/skills/audio_tts/scripts/models/Qwen3-TTS-12Hz-0.6B-CustomVoice \
"Hello, this is a test." \
Vivian \
english
ls -la output.wav
```
## Manual Installation
If automatic download fails, manually download the components:
1. Go to https://github.com/second-state/qwen3_tts_rs/releases/latest
2. Download the zip for your platform:
- `qwen3-tts-linux-x86_64.zip` (Linux x86_64 CPU, includes libtorch)
- `qwen3-tts-linux-x86_64-cuda.zip` (Linux x86_64 CUDA, includes libtorch)
- `qwen3-tts-linux-aarch64.zip` (Linux ARM64 CPU, includes libtorch)
- `qwen3-tts-linux-aarch64-cuda.zip` (Linux ARM64 CUDA / Jetson, includes libtorch)
- `qwen3-tts-macos-aarch64.zip` (macOS Apple Silicon, includes mlx.metallib)
3. Extract to `~/.openclaw/skills/audio_tts/scripts/`:
```bash
unzip qwen3-tts-*.zip -d ~/.openclaw/skills/audio_tts/scripts/
```
4. Make executable:
```bash
chmod +x ~/.openclaw/skills/audio_tts/scripts/tts
chmod +x ~/.openclaw/skills/audio_tts/scripts/voice_clone
```
7. Copy reference audio files from the repo into the scripts directory:
```bash
git clone --depth 1 https://github.com/second-state/qwen3_tts_rs.git /tmp/qwen3-tts-repo
cp -r /tmp/qwen3-tts-repo/reference_audio ~/.openclaw/skills/audio_tts/scripts/reference_audio
rm -rf /tmp/qwen3-tts-repo
```
8. Download the TTS models:
```bash
pip install huggingface_hub transformers
MODELS_DIR=~/.openclaw/skills/audio_tts/scripts/models
mkdir -p "$MODELS_DIR"
huggingface-cli download Qwen/Qwen3-TTS-12Hz-0.6B-CustomVoice --local-dir "$MODELS_DIR/Qwen3-TTS-12Hz-0.6B-CustomVoice"
huggingface-cli download Qwen/Qwen3-TTS-12Hz-0.6B-Base --local-dir "$MODELS_DIR/Qwen3-TTS-12Hz-0.6B-Base"
```
9. Generate tokenizer.json files for each model:
```bash
python3 -c "
from transformers import AutoTokenizer
for model in ['Qwen3-TTS-12Hz-0.6B-CustomVoice', 'Qwen3-TTS-12Hz-0.6B-Base']:
path = '$MODELS_DIR/' + model
tok = AutoTokenizer.from_pretrained(path, trust_remote_code=True)
tok.backend_tokenizer.save(path + '/tokenizer.json')
print(f'Saved {path}/tokenizer.json')
"
```
## Troubleshooting
### Download Failed
Check network connectivity:
```bash
curl -I "https://github.com/second-state/qwen3_tts_rs/releases/latest"
```
### Unsupported Platform
Check your platform:
```bash
echo "OS: $(uname -s), Arch: $(uname -m)"
```
Supported: Linux (x86_64, aarch64) and macOS (Apple Silicon arm64).
### Missing libtorch (Linux only)
Ensure `LD_LIBRARY_PATH` includes the libtorch lib directory. The SKILL.md instructions set this automatically.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |