Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Converts text to speech using SiliconFlow's CosyVoice2 model. Offers 8 preset voices across 4 languages and 5 Chinese dialects, with 150ms first-packet latency and voice cloning from 3-second samples.
CosyVoice2 via SiliconFlow provides native Chinese dialect support that most TTS APIs lack, including Cantonese, Sichuan, and Shanghainese.
When to use it
- Generating voiceovers for short videos in Mandarin or Cantonese
- Creating audio versions of written content in multiple languages
- Prototyping voice UI responses with different character tones
- Batch-generating narration files for e-learning materials
- Cloning a specific voice for consistent branded audio output
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: generate an MP3 of 'Welcome to our service' in a gentle female voice at 0.9x speed
- 1Reads SILICONFLOW_API_KEY from environment
- 2Calls SiliconFlow TTS API with voice set to claire and speed 0.9
- 3Receives audio stream with ~150ms first-packet delay
- 4Writes response to output.mp3 locally
output.mp3 saved to current directory with claire voice at 0.9x speed
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.