Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Adds a voice interface to any OpenClaw agent. Browser mic audio streams through a configurable STT provider to your OpenClaw gateway, and the response comes back as spoken audio through a configurable TTS provider. Targets sub-2 second time-to-first-audio with full streaming at every stage.
Full end-to-end streaming across audio capture, STT, LLM, and TTS keeps total latency under 2.5 seconds, which browser-based voice tools typically don't achieve.
When to use it
- Dictating tasks to your agent hands-free while cooking or commuting
- Running voice Q&A with a document-aware agent
- Talking through code problems out loud and hearing solutions back
- Getting spoken summaries of long reports without reading
- Using multilingual voice input via ElevenLabs STT
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: "Set up voice chat with Deepgram STT and ElevenLabs TTS"
- 1Copies .env.example to .env in the skill directory
- 2Sets VOICE_STT_PROVIDER=deepgram and VOICE_TTS_PROVIDER=elevenlabs in .env
- 3Prompts user to add DEEPGRAM_API_KEY and ELEVENLABS_API_KEY
- 4Runs uv run scripts/server.py to start the voice server
- 5Opens http://localhost:7860 in the browser
Voice server running on port 7860. Click the mic to start talking to your agent.
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.