How romerik/voice-control-unitree-go2 works
Project type and purpose
• This is an open-source Python application that provides a realtime, multimodal “copilot” experience using Chainlit.
• Key features:
• Real-time Python client for the OpenAI realtime API
• Multimodal I/O: speech and text in parallel
• Tool-calling framework for executing tasks
• Gesture control integration (e.g., Unitree Go2 robot)
• Internationalized UI with Chainlit translations
File structure
• Top-level files:
• app.py / main.py / main_app.py – entry points for different modes (Chainlit UI, CLI, gesture launcher)
• gesture-receiver.py, launch-gesture-control.py – modules to capture and forward gesture events
• robot_control.py – interfaces to control the Unitree Go2 robot
• realtime/ – contains realtime API client wrappers and custom tool definitions
• .chainlit/ – Chainlit configuration (config.toml) and translations for UI
• requirements.txt – lists dependencies (openai-realtime, chainlit, etc.)
• Patterns:
• Modular separation: UI (app.py), API layer (realtime/), tool layer (realtime/tools.py), hardware layer (robot_control.py and gesture modules)
• Configurations for environment (.env.example) and Chainlit UI
README for additional insights
• Emphasizes realtime API, multimodal I/O, tool calling, visual presence cues
• Mentions Chainlit cookbook context
Final reminder
• Essential interactions and data flows without overcomplicating.