Google Meet’s speech translation now works in 63 languages on the web, Android and iOS. Powered by Gemini 3.5 Live Translate, it detects which language each person is speaking. It then translates their speech into each listener’s own language.
The feature is paid-only. You need Google AI Pro or Ultra, or a selected Workspace plan. Lower tiers also cap it at 90 minutes per meeting.
How Google Meet speech translation works now
Until now, Meet supported a limited set of language pairs, with one pair set for the whole meeting. The new version detects languages automatically instead.
Each participant turns translation on for themselves from the meeting toolbar. You can also mark languages you already understand, so Meet leaves those untranslated. Captions match the translated audio.
Meet hardware gets listening support. People in a meeting room can hear translated speech from remote participants. Speech from the room, however, isn’t translated for others.
Which plans get it
- Workspace Business: Business Standard and Business Plus
- Workspace Enterprise: Enterprise Standard and Enterprise Plus
- Frontline: Frontline Plus
- Consumer: Google AI Pro and Google AI Ultra
- Education: Google AI Pro for Education
- Add-ons: AI Expanded Access and AI Ultra Access
Everyone in a meeting organized by an eligible user can use it, including guests from other domains. In one-to-one calls, both people get it if either has an eligible plan.
Meeting length is capped by plan. Eligible base Workspace plans and Google AI Pro allow up to 90 minutes per meeting. AI Expanded Access, AI Ultra Access and Google AI Ultra raise that to 8 hours.
That means free Google account users can’t start a translated meeting themselves. Google AI Pro subscribers who pay for access still hit a 90-minute ceiling, which a long workshop or interview can easily pass.
Where it doesn’t work
Translation isn’t available in live streams, meeting recordings or client-side encrypted meetings. It also doesn’t work on dial-in calls, third-party interoperability calls, or with adaptive audio. Google adds that availability is subject to regional and age restrictions.
Privacy and admin defaults
Google makes one privacy promise about the feature in its announcement:
“Google does not store voice data or train models on your voice.”
That’s Google’s own claim, not something androidpure has independently verified. Users choose whether their speech is translated, and can change that in Meet settings.
Business readers should check their defaults. For eligible Workspace editions with Gemini for Meet enabled, the feature is on by default at the admin level. Your organization may already have it switched on. For end users, translation is off at the start of each meeting.
Rollout timeline
Rapid Release domains started getting it on October 22, with rollout taking one to three days. Scheduled Release domains follow from November 2, over up to 15 days. On Android and iOS, update the Meet app to get it.
Meet has been adding other language tools on mobile too, including interpreter tile pairing on Android and iOS.
Source: Google Workspace Updates blog
