gradio-app/fastrtc

Supports bidirectional audio and video streaming over WebRTC and WebSockets through FastAPI integration, with built-in voice activity detection and automatic turn-taking logic. Includes optional text-to-speech and speech-to-text capabilities, plus launching modes via Gradio UI, temporary phone numbers, or custom frontends. Targets AI/ML applications by integrating with LLM APIs (OpenAI, Anthropic, Google Gemini) and supporting real-time multimodal interactions.

4,547 stars and 51,336 monthly downloads. Available on PyPI.