OpenClawSkills
GitHub
Use Case • TutorialHeader.readTime

Voice Assistant Setup

Create a complete hands-free voice assistant workflow with Voice Wake and Talk Mode.

Transform OpenClaw into an always-listening, conversational voice assistant similar to Jarvis or Alexa, but with the power of LLMs and full control over your tools.

Tutorial.step

What You'll Build

  • Fully hands-free Voice Wake activation
  • Low-latency Talk Mode for natural conversations
  • Voice-triggered automation workflows
  • Integration with Smart Home, Calendar, and Mail
Tutorial.step

Voice System Components

To get the best experience, we use two core features:

  • Voice Wake: Uses local wakeword detection (e.g., "Hey Claw") so the agent only listens when called.
  • Talk Mode: A continuous audio stream that uses ElevenLabs for high-quality, expressive voice responses.
Tutorial.step

Step 1: Enable Voice Wake

Configuration for local wakeword detection:

openclaw.json
{
  "messages": {
    "tts": {
      "wake": {
        "enabled": true,
        "word": "Claw",
        "sensitivity": 0.5
      }
    }
  }
}
Tutorial.step

Step 2: Configure Talk Mode

Connect ElevenLabs for low-latency text-to-speech:

openclaw.json
{
  "messages": {
    "tts": {
      "talk": {
        "enabled": true,
        "provider": "elevenlabs",
        "voiceId": "your-voice-id-here"
      }
    }
  }
}
Tutorial.step

Step 3: Voice-Only Commands

Once setup, you can perform tasks without ever typing:

Task Management

"Claw, add 'buy milk' to my grocery list and remind me when I'm home."

Information

"Claw, summarize the top three stories on Hacker News right now."

Tutorial.step

Advanced Multi-Language

Continuous Conversation

Talk Mode supports multi-turn conversations. You can ask follow-up questions or have a debate without re-triggering the wake word every time.