Reachy Talks: Voice-Activated Conversation #2
Description
Overview
Voice-activated conversation with LLM integration. Ask Reachy questions and get animated responses.
Features
- Wake word detection ("Hey Reachy")
- Speech-to-text for user input
- LLM for response generation (OpenAI/Claude/local)
- Text-to-speech with expression synchronization
- Animated responses: head movements, antenna reactions, expressions
Technical Approach
- Use faster-whisper for local STT
- TTS options: OpenAI, ElevenLabs, or local piper
- Robot expressions tied to speech sentiment
- Interrupt handling (stop talking when user speaks)
Integration
- Build on conversation_app foundation
- Add personality system (mood, memory)
- Custom voices for different personalities
This is a roadmap item tracked at runreachyrun.com