Voice in Chat

Voice in Chat is conversational speech: start a realtime voice session, talk, and keep the transcript. It runs on Ethen Voice, the platform for speech and voice AI that also powers audio in Studio. There is no separate Voice app to install, no telephony product, and no voice marketplace.

How a session starts

1. You must be signed in — anonymous voice is rejected.
2. Chat mints a short-lived, single-use realtime token from the server. Your browser receives only that token and the gateway WebSocket URL; the long-lived provider key never leaves the server.
3. The session opens over the realtime protocol with rate limits applied. Malformed session requests fail explicitly.

If voice is not configured on the deployment, session creation answers 503 VOICE_NOT_CONFIGURED and names the missing server key. That is a deployment state, not a bug in your request.

Transcripts

Sessions produce transcripts you can read after the call. Transcript handling follows the same data rules as chat history — scoped to your account.

Creation voice lives in Studio

Speech you create as media — cloned voices, dubbing, audio projects — belongs to Studio voices. Chat voice is for talking with Ethen; Studio voice is for making audio with Ethen.

Next steps

• Chat overview
• Conversations
• Studio voices