Voice

Real-time speech-to-speech translation

Speak and hear the translation in near-real time. Pick your languages, choose interactive or simultaneous mode, and Voice bridges you to a live translation model over WebSocket.

What you can do

Interactive mode

Sub-second translation that's best for a two-way conversation — the model listens as you speak.

Simultaneous mode

Continuous, always-on translation that's ideal for lectures and meetings.

Live transcripts

Both your words and the translation appear on screen as they happen.

Custom glossary

Pin terms to their exact translation so names and jargon stay consistent.

20+ languages

Translate between a practical set of supported languages, or let it auto-detect.

How it works

Step 1

Pick languages

Choose source & target, plus a mode.

Step 2

Press start

The session opens a live translation stream.

Step 3

Just speak

Hear the translation in near-real time.

Cost

40 credits per minute

Charged per started minute of live conversation. Top up to keep the session going.

Get started