AI Voice Agents
Home / Realtime frameworks

Moshi

Full-duplex speech-text model from Kyutai

Moshi is an open-source speech-text foundation model from Kyutai for realtime voice conversation. It listens and speaks at the same time, which allows full-duplex dialogue.

realtimespeech-modelfull-duplexopen-weights
Moshi alternatives

More in Realtime frameworks

Featured

LiveKit Agents

Realtime frameworks

Open-source framework for realtime voice agents

Open sourcerealtimewebrtc

Agora

Realtime frameworks

Realtime voice and video SDKs with AI agents

Freemiumrealtimesdk

Bolna

Realtime frameworks

Open-source framework for phone voice agents

Open sourcerealtimevoice-agents

Daily

Realtime frameworks

WebRTC APIs and the company behind Pipecat

Freemiumwebrtcapi

FastRTC

Realtime frameworks

Python library for realtime audio apps

Open sourcerealtimewebrtc

LiveKit

Realtime frameworks

Open-source WebRTC platform with a hosted cloud

Freemiumwebrtcsip