The interface your Mac
was waiting for.
ClawFace brings native, on-device voice intelligence to OpenClaw. Powered by local Whisper models, Active Listening, and a real-time Markdown engine.
Core Capabilities
Native Whisper
On-device transcription (tiny to large-v3). Zero latency, full privacy. No cloud APIs required for speech-to-text.
Active Listening
Continuous hands-free loop. Advanced VAD detects when you speak and when you stop. Just talk, it listens.
Rich Chat Engine
Streaming markdown, syntax highlighting, multimedia attachments, and deduplication. A modern chat UX built for devs.
Technical Specifications
Audio Intelligence
- • Rust-based native recording via cpal.
- • Advanced VAD: Zero-crossing, Crest factor, Spectral analysis.
- • Real-time Energy: Visual feedback loop.
- • 4-State FSM: Idle → MaybeSpeech → Speaking → Silence.
OpenClaw Protocol
- • WebSocket Bidirectional: Full handshake & auto-reconnect.
- • Delta Streaming: Token-by-token visual updates.
- • Context Aware: Live token usage dashboard (Red/Yellow/Green).
- • Tool Invocation: Direct HTTP hooks for agent actions.
Live Dashboard
Monitor agent health, active cron jobs, next run times, and model context usage in real-time.
System Tray
3 dynamic states (Listening, Processing, Speaking). Quick toggles for models and voice modes.
Rich Multimedia Support
Base64 encoded client-side for privacy. Syntax highlighting for all major languages.