Pipecat

Pipecat is an open-source Python framework for building real-time voice AI applications — voice assistants and conversational agents that need low-latency audio input/output alongside LLM processing.

It handles the plumbing of a real-time voice pipeline (audio streaming, speech-to-text, LLM inference, text-to-speech, interruption handling) so builders focus on the conversation logic rather than the real-time audio infrastructure underneath.

Used for building voice agents (phone-based AI assistants, voice-controlled apps) where response latency matters enough that a naive request-response API pattern isn't fast enough.