← Back to all projects

Pipecat

Real-time voice and multimodal agent pipelines

OfficialBSD-2-Clause
Stars
16.9k
Forks
3k
Open issues
495
Last commit
29 Oct 2026

What is Pipecat?

Assembles the whole speech-to-speech loop — transcription, model, synthesis, interruption handling — as a pipeline of swappable services, which is the tedious part to get right. Latency is the entire game in voice, and turn-taking and barge-in are first-class here rather than bolted on. Running it well means caring about transport and network topology, not only which model you picked.

What can you do with Pipecat?

  • Scaffold and deploy from the CLI — The CLI ships with the package via uv tool install "pipecat-ai[cli]"; pipecat init quickstart scaffolds a bot that runs, and the same CLI monitors and deploys it to production.
  • Compose a multi-agent system — Each pipeline counts as one agent, and pipelines hand off to each other, fan out in parallel, or run as sidecar workers over a shared bus, in one process or spread across machines.
  • Script conversation paths with Flows — Pipecat Flows is built into the framework for predefined or dynamic conversation paths with state management, with working versions under examples/flows.
  • Connect clients on any platform — Official client SDKs cover JavaScript, React, React Native, Swift, Kotlin and C++, plus an ESP32 build for hardware, connecting over WebSocket or WebRTC transports.
  • Debug a live pipeline — Whisker attaches to a running pipeline as a real-time debugger for its processors, and Tail presents the same activity as a terminal dashboard.

Star history

15 Aug to 30 Oct · +2.8k

14.1k16.9k

Frequently asked questions

Is Pipecat free for commercial use?

Pipecat is released under the BSD-2-Clause licence — OSI-approved open source, which permits commercial use.

How can Pipecat be deployed?

Pipecat is available as Self-hosted / Runs locally / Managed cloud.