Pipecat

Pipecat is an open-source Python framework for building real-time voice and multimodal conversational agents. It orchestrates audio, video, AI services, and transport layers into composable pipelines, letting developers focus on what makes their agent unique rather than infrastructure plumbing. Licensed under BSD 2-Clause, Pipecat supports a wide ecosystem of speech, LLM, and transport integrations out of the box.

• Voice-first pipeline architecture — install via pip install pipecat-ai or uv add pipecat-ai and wire together STT, LLM, and TTS services in minutes
• Extensive service integrations — supports 20+ STT providers (Deepgram, AssemblyAI, OpenAI Whisper, Azure, Google, etc.), 20+ LLMs (OpenAI, Anthropic, Gemini, Groq, Mistral, etc.), and 30+ TTS engines (ElevenLabs, Cartesia, AWS, Azure, etc.)
• Real-time transport options — connect via WebRTC (Daily, LiveKit), WebSockets, FastAPI, or WhatsApp for ultra-low latency interactions
• Speech-to-speech support — integrates OpenAI Realtime, Gemini Multimodal Live, AWS Nova Sonic, and Ultravox for end-to-end voice pipelines
• Multi-agent systems — use Pipecat Subagents to build distributed agents that communicate through a shared message bus and hand off conversations between specialists
• Structured conversations — Pipecat Flows manages complex conversational states and transitions for guided dialog systems
• Client SDKs for every platform — official SDKs for JavaScript, React, React Native, Swift (iOS), Kotlin (Android), C++, and ESP32
• Audio processing utilities — built-in VAD (Silero), noise cancellation (Krisp Viva, RNNoise, Koala), and audio filtering
• Observability and analytics — integrates OpenTelemetry and Sentry for pipeline monitoring and error tracking
• CLI tooling — use pipecat init quickstart or the Pipecat CLI to scaffold, monitor, and deploy agents to production
• Debugging tools — Whisker provides real-time pipeline debugging; Tail offers a terminal dashboard for live monitoring
• Community integrations — browse and contribute service integrations via the community integrations registry