voice-ui

Voice UI

Self-evolving AI assistant UI - Can improve itself while conversing by voice

Features

• 🎤 Voice recognition (Whisper)
• 🔊 Speech synthesis (TTS)
• 🤖 Cute robot UI (facial expression changes)
• 🔄 Self-evolution (instruct UI changes by voice)
• 📝 Automatic Git commits

Setup

cd <workspace>/skills/voice-ui
npm install
./start.sh

In the browser http://localhost:8765 Open

Required setup

OpenClaw config (~/.openclaw/openclaw.json) add a voice agent:

{
  "agents": {
    "list": [{
      "id": "voice",
      "name": "Voice Assistant",
      "model": { "primary": "anthropic/claude-sonnet-4-5" }
    }]
  }
}

How to use

1. Tap the microphone button (or hold down the space key)
2. Speak to it
3. The AI responds

Self-evolution commands

• "Make the background blue" → automatically edits CSS
• "Make the buttons bigger" → changes the styles
• "Add a new feature" → edits the JS

Changes are automatically committed to Git.

File structure

• index.html - Main UI
• server.cjs - Node.js server
• start.sh - Startup script
• CONTEXT.md - Context information for the AI

Environment variables

• OPENAI_API_KEY - OpenAI API Key (for Whisper/TTS)

If there is no configuration, it tries to retrieve it automatically from OpenClaw's settings.