voice-tts

What This Skill Does

The voice-tts skill integrates Microsoft's edge-tts engine into the OpenClaw ecosystem, enabling the agent to synthesize high-quality, human-like Chinese speech from text. This skill transforms the agent from a purely text-based interface into a multimodal assistant capable of delivering audio responses. Whether you are using Feishu, Telegram, or Discord, this skill processes text input to generate opus-format audio files, which are then transmitted as native voice messages. It supports a diverse range of voices, including various gender-specific neural voices and regional accents like Northeast Mandarin and Taiwanese Mandarin, providing a professional and localized communication experience.

Installation

To use this skill, ensure you have the edge-tts Python library installed. We strongly recommend using pipx to manage the installation to keep your system Python environment clean and avoid dependency conflicts.

1. Install via pipx install edge-tts (or pip install --user edge-tts for Linux systems where pipx is unavailable).
2. Verify the installation by running edge-tts --list-voices in your terminal.
3. Ensure the script directory is accessible by the OpenClaw agent to allow for the automated triggering of tts.sh.

Use Cases

• Proactive Notifications: Automatically send voice reminders for meetings or project deadlines in team chat channels like Feishu.
• Customer Service Automation: Provide personalized, warm human-sounding voice responses in Telegram or Discord customer support threads.
• Accessibility Enhancements: Convert long-form text reports into audio briefs for users who prefer listening to content while on the move.
• Regional Localization: Use specific dialect voices (e.g., zh-TW-HsiaoChenNeural) to better engage with specific demographic user bases.

Example Prompts

1. "Send a voice message to the team channel saying 'The project review meeting will start in 5 minutes, please be prepared.'"
2. "Reply to this user on Telegram with an audio message: 'Thank you for your feedback, we have received your request and will address it shortly.'"
3. "Use the Xiaoxiao voice to announce this text in Discord: 'Welcome everyone, today's focus is on optimizing our AI agent workflows.'"

Tips & Limitations

• Optimization: Keep your input text between 50-300 characters for the best balance of generation speed and audio quality.
• File Management: Note that the skill generates files in ~/.openclaw/media/. Regularly clean up these directories to prevent disk bloat if you generate high volumes of audio.
• Network Dependency: While edge-tts is locally executed, it requires an active internet connection to communicate with Microsoft's speech services during the generation phase.
• Latency: Generation typically takes 1-3 seconds. Plan your application flow to account for this short processing delay before sending the message.