What is Pipecat?
Pipecat is an open-source framework for building real-time voice and multimodal AI agents, developed by Daily. It orchestrates the stages of a voice agent pipeline - ASR, LLM, TTS, enhancement, and VAD.
What is an example of Pipecat in use?
A developer building an outbound sales voice agent wires together Deepgram for ASR, GPT-4 for reasoning, and ElevenLabs for TTS using Pipecat, then deploys it through a telephony provider like Twilio.
How does Pipecat work?
Pipecat uses a frame-based pipeline model: audio, text, and other modalities flow as frames through chained processors. It includes integrations with popular ASR, LLM, TTS, and telephony providers, plus primitives for VAD, interruption handling, and tool calling.
How does ai-coustics work in Pipecat?
ai-coustics integrates with Pipecat as a real-time enhancement stage. Voice agents built on Pipecat can route incoming audio through Quail before it reaches ASR, ensuring accurate transcription even on noisy phone calls and through Quail VAD and Quail Voice Focus where turn-taking precision or speaker isolation matters.
