WebRTC + OpenAI Realtime
Realtime 4o Voice
A minimal, browser-native voice UI powered by WebRTC and the OpenAI Realtime API.
Disconnected
Choose before connecting. Voice can’t change after audio starts.
Sets transcription language and nudges the assistant’s reply language.
Enable MCP tools
The OpenAI Docs MCP is read-only. Add your own server URL to query other MCP tools.
Transcript (best effort)
No turns yet.