LLM Voice Biometrics for Home Assistant
Local speaker identification. Zero cloud. Intelligence over convenience.
Next-generation voice biometrics powered by Sherpa-ONNX
Lightning-fast speaker recognition with neural network embeddings. Generate embeddings in under 100ms—60-162x faster than WeSpeaker.
Sophisticated identity management with transform, merge, and alias operations. Handle complex family dynamics and voice variations seamlessly.
Deploy passive nodes per room with RTSP audio capture. Identify speakers across your entire home with room-specific confidence scoring.
100% local processing. No cloud dependencies. Your voice data never leaves your network. Full control over enrollment and blocklists.
Seamless integration with conversation.process and STT pipelines. Speaker tags appear naturally in transcripts for context-aware AI responses.
Full MQTT discovery. Per-person devices with blocklist controls. Global and per-room confidence sensors. Dashboard-ready enrollment interface.
Beautiful Home Assistant integration
Web-based enrollment interface with real-time active pipeline monitoring and blocklist management.
Per-person and per-room devices with confidence sensors, voice detection, and threshold controls.
Watch speaker identification happen in real-time across multiple rooms with confidence scoring.
Get VoiceBM running in under 10 minutes
Run the automated installer wizard to configure your environment.
Install the 10 global services and start the core system.
Deploy passive nodes for each room you want to monitor.
Download the complete package and get started
Size: 154KB | SHA256: 83db672d0761eee6100f7d3c8f569df99b540277925f73e4c42c9381d537de74
Includes: All templates, deployment scripts, documentation, and setup wizard
Both AI assistants have complete VoiceBM documentation access
Custom GPT with all VoiceBM documentation pre-loaded. Click and start chatting immediately—no setup required. Perfect for quick questions and troubleshooting.
💬 Chat with GPT InstantlyCo-architect of VoiceBM. Quick 3-step setup loads all documentation automatically. Perfect for advanced customization, design philosophy, and complex questions.
🚀 Setup Claude SupportGPT: Instant access, pre-configured | Claude: 30-second setup, full flexibility