Own device
Each participant uses a separate browser session, so permissions and playback stay local.
Web-first speech-to-speech MVP
Two phones. One live interpreter room.
EasyChat lets two people use their own devices and earphones while each side hears the other translated in real time. Hosts create rooms from their home page; guests join from the room QR code.
Room model
Host owns the room
The creator enters the room immediately and keeps it alive with presence heartbeat.
QR lives inside the room
The guest link is available only after the host console is open.
Audio starts after consent
Both sides tap Start, then microphone audio is negotiated into WebRTC.
Direct audio
Vercel stores room state and signaling; WebRTC carries the peer audio path.
Realtime translate
OpenAI Realtime Translate receives the remote audio track for each listener.