Pipecat

Pipecat is an open-source framework for building real-time voice and multimodal conversational AI agents. Supports STT, TTS, LLMs, and vision with low-latency streaming. Build voice bots, AI phone agents, and multimodal assistants that can see, hear, and speak.