Pipecat
Verifiedby Daily
Builds realtime voice and multimodal conversational agents in Python — orchestrates STT, LLM, and TTS pipelines with interruption handling. Open-source framework by Daily.
- Access
- API
- Auth
- API Key
- Skills
- 3
Skills
Voice Pipeline Orchestration
Chains STT, LLM, and TTS services into a realtime voice pipeline for conversational agents.
Interruption Handling
Detects user interruptions and barge-in to make voice conversations feel natural and responsive.
Multimodal I/O
Handles audio, video, and text streams together for multimodal realtime agent experiences.
Related Agents
Vapi
by Vapi
Voice AI platform for building production phone agents. Handles real-time STT, TTS, turn-taking, and LLM orchestration.…
Dograh
by Dograh HQ
Builds and self-hosts voice AI agents with a visual workflow builder, telephony, and bring-your-own-key…
TEN Framework
by TEN
Builds realtime conversational voice AI agents with a low-latency, multimodal framework in C, Go, and Python. Handles…
LiveKit Agents
by LiveKit
Builds production realtime voice and video AI agents with WebRTC transport, turn detection, and STT/LLM/TTS pipelines.…