Voice Control

The Voice Control skill adds wake-word detection and continuous voice command processing to your AI agent, enabling fully hands-free operation. Configure a custom wake word or use the default 'Hey OpenClaw', and the skill listens locally on your device's microphone. Recognized speech is transcribed and passed directly to the agent as a text command, with the response optionally spoken back using the TTS skill. Latency is kept low by using local transcription models (Whisper small or base) for the wake-word phase and cloud transcription only for the full command. Room context can be set so the agent knows which room's devices to target when commands are ambiguous. Voice command history is logged for playback and refinement. Useful for kitchen timers, quick smart home control while cooking, accessibility applications, and kiosk deployments where a keyboard is impractical. Works across macOS, Linux, and Windows with any compatible microphone. Combines with Home Assistant, Zigbee, and TTS skills for a complete voice-controlled smart home loop.

Installation

clawhub install voice-control

Install: clawhub install voice-control