yusufferdogan/openclaw-qwen-tts-skill

Overview

This skill provides a lightweight command-line toolkit for generating multilingual CustomVoice samples and one-off voice replies using Qwen3‑TTS. It includes a Conda Python 3.12 environment setup, sample batch generation, simple CLI scripts for single replies, and an optional auto-voice-reply toggle. No Remotion or video rendering is included — audio-only workflows only.

How this skill works

The package installs a Python 3.12 Conda environment and qwen-tts, plus system dependencies like FFmpeg. Shell scripts wrap Python code to generate WAV files in a tmp directory for batch samples or immediate single replies, returning the path to the generated file. A small toggle mechanism enables or disables automatic voice replies so you can switch between text-only and voice-enabled outputs.

When to use it

• Generate multilingual TTS samples quickly from short scripts or CI jobs.
• Produce one-off voice replies for chatbots, notifications, or demos.
• Test different Qwen3‑TTS speakers or languages without a heavy deployment.
• Create WAV assets for prototyping audio UX or accessibility features.
• Run local experiments where GPU acceleration is available for faster synthesis.

Best practices

• Run the provided setup script to create the Conda Python 3.12 environment before use.
• Install FFmpeg via Homebrew on macOS to ensure correct audio encoding and trimming.
• Store generated WAV files in the designated tmp folder and clean periodically.
• Switch models in the Python script only when you need higher quality or cloning features.
• Use GPU where available to reduce generation time; CPU works for small batches.

Example use cases

• Batch-generate a set of voice samples for UI voice selection testing.
• Produce a quick WAV reply for a chat interface using the one-off voice script.
• Enable auto-voice replies for a local demo, then disable them for logging runs.
• Compare different speakers and languages by swapping SPEAKER and LANG env vars.
• Integrate the CLI scripts into small automation pipelines for audio asset creation.