Voice Commands and Talk Mode with OpenClaw

Voice Features

OpenClaw supports voice interaction through two modes: Voice Wake for quick commands and Talk Mode for continuous conversation.

Voice Wake

Voice Wake is always-on wake word detection. Say "Hey OpenClaw" and the system activates, listens for your request, and responds audibly.

Supported Platforms

• macOS (native)
• iOS (via companion app)
• Android (via companion app)

Setup

Enable Voice Wake in your configuration:

{
  "voice": {
    "enabled": true,
    "wake_word": "hey openclaw",
    "tts_provider": "elevenlabs"
  }
}

Usage

1. Say "Hey OpenClaw"
2. Wait for the activation chime
3. Speak your request naturally
4. OpenClaw responds audibly

Talk Mode

Talk Mode enables continuous hands-free conversation. Instead of waiting for a wake word each time, you enter a back-and-forth dialogue.

Starting Talk Mode

• Say "Hey OpenClaw, let's talk"
• Or use the keyboard shortcut (configurable)

Features

• Interruption detection: Break in when OpenClaw is speaking
• Natural pauses: OpenClaw detects when you're done talking
• Context continuity: Same conversation context as text chat
• Exit: Say "stop talking" or "goodbye"

Text-to-Speech (TTS)

OpenClaw uses ElevenLabs for natural-sounding voice responses.

Configuration

{
  "voice": {
    "tts_provider": "elevenlabs",
    "tts_voice": "rachel",
    "tts_speed": 1.0
  }
}

Voice Options

ElevenLabs offers multiple voices with different characteristics. Choose one that suits your preference.

Speech-to-Text (STT)

Incoming voice is transcribed using Whisper or similar:

{
  "voice": {
    "stt_provider": "whisper",
    "stt_language": "en"
  }
}

Use Cases

• Hands-free cooking: "Hey OpenClaw, what's the next step in my recipe?"
• Driving: "Hey OpenClaw, add milk to my shopping list"
• Exercise: "Hey OpenClaw, start a 5-minute timer"
• Accessibility: Full AI assistance without a keyboard

Tips

• Speak clearly and at a normal pace
• Pause briefly after the wake word
• Background noise can affect recognition — quiet environments work best
• Voice messages on WhatsApp/Telegram are also transcribed automatically