An open-source real-time digital-human pipeline that combines LLM replies, speech recognition and synthesis, WebRTC playback, and pluggable avatar models for privately deployed conversational characters.