Setup & Installation
openclaw skills install @jithinm/qwen3-tts-local-inference
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
npx clawhub install qwen3-tts-local-inference
What This Skill Does
Runs Qwen3-TTS speech synthesis locally via Python with no server or API required. Supports three modes: built-in speakers, natural language voice description, and voice cloning from a short audio clip. Outputs .wav files and works fully offline after the initial model download.
Runs entirely offline after setup, avoiding API rate limits, per-character costs, and data privacy concerns tied to cloud TTS services.
When to use it
- Generating voiceovers for video content without cloud costs
- Cloning a speaker's voice from a 3-second audio clip
- Producing multilingual audio narration in Chinese, English, Japanese, or Korean
- Embedding TTS generation directly into a Python data pipeline
- Quickly prototyping spoken UI with a specific emotional tone or accent
Example Workflow
Here's how your AI assistant might use this skill in practice.
INPUT
User asks: clone my voice from a short recording and read back a script in English
AGENT
- 1Locate the reference audio file (~3 seconds) and its transcript
- 2Run tts.py with --mode voice-clone, --ref-audio, --ref-text, and the target text
- 3Monitor output for the generated .wav file path
- 4Confirm duration and inference time from the JSON result
OUTPUT
A .wav file in tts_output/ spoken in the cloned voice