FastRTC

FastRTC is a Python library that enables developers to transform any Python function into real-time audio and video streams over WebRTC or WebSockets. Built by the Gradio team, it provides a simple yet powerful interface for creating voice-enabled applications, video processing pipelines, and interactive communication tools with minimal code.

The library offers three main deployment methods: a built-in UI for testing and sharing streams, a fastphone feature for temporary phone number access, and FastAPI integration for production systems.

• Stream Class provides the core abstraction for handling real-time communication, supporting audio, video, and combined audio-video modalities with send-receive or receive-only modes.

• Automatic Voice Detection includes built-in pause detection and turn-taking capabilities through the ReplyOnPause handler, allowing developers to focus on response logic rather than audio processing mechanics.

• Built-in UI launches a Gradio-powered interface with .ui.launch() for rapid prototyping and testing of streams without building custom frontends.

• WebRTC and WebSocket Support enables mounting streams on FastAPI applications to expose endpoints for custom frontend integrations with full real-time communication capabilities.

• Telephone Integration offers the .fastphone() method to obtain free temporary phone numbers for calling into streams, requiring only a Hugging Face token.

• Speech-to-Text and Text-to-Speech plugins are available through optional extras (vad, stt, tts), enabling voice-to-voice AI applications with LLM integration.

• Video Processing supports real-time video manipulation including object detection, image transformations, and webcam streaming with customizable handlers.

• Plugin Ecosystem provides galleries for text-to-speech, speech-to-text, and turn-taking models, making it easy to swap components for different use cases.

To get started, install with pip install fastrtc and optionally add extras for voice features. Create a Stream with a handler function, then launch the UI or mount on FastAPI. The library handles WebRTC negotiation, audio chunking, and stream management automatically.