Google has expanded real-time speech translation in Google Meet to 63 languages [1]. In an announcement on 22 October 2026, Google said Meet now detects the spoken language automatically. Each participant now turns translation on for themselves, instead of one language pair being set for the whole meeting [1].
The update reaches Rapid Release domains from 22 October 2026 and Scheduled Release domains from 2 November 2026 [1]. On eligible base Workspace plans and Google AI Pro, translation runs for up to 90 minutes per meeting [1]. The details below come from the announcement [1].
What changes in real-time speech translation in video meetings?
Until now, speech translation in Meet supported a limited set of language pairs. One pair had to be configured for the entire meeting [1]. Google says the new version runs on Gemini 3.5 Live Translate. It translates between any combination of the 63 supported languages in the same meeting [1].
Three changes stand out:
- Automatic detection. Meet detects which language is being spoken. It translates that speech into each listener's preferred language in near real time. This works even when several languages are spoken or a speaker switches language [1].
- Personal controls. Each participant turns translation on with a control in the meeting toolbar. They can also list languages they already understand, so those are not translated [1].
- Matching captions. When translation is on, the captions match the translated audio [1].
Translation works on the web, Android and iOS [1]. People in rooms with Google Meet hardware can listen to translated speech from remote participants. However, Google says "speech originating from the conference room hardware will not be translated for others" [1].
Who can use it, and for how long?
If the meeting organiser has an eligible plan, every participant can use translation, including guests from other domains [1]. In direct calls between two people, both can use it if either person has an eligible plan [1].
The time limit depends on the plan [1]:
- Eligible base Workspace plans and Google AI Pro: up to 90 minutes per meeting [1].
- AI Expanded Access, AI Ultra Access and Google AI Ultra: up to 8 hours per meeting [1].
Eligible plans are Business Standard and Plus, and Enterprise Standard and Plus. Google AI Pro and Ultra, Frontline Plus and Google AI Pro for Education also qualify, as do the AI Expanded Access and AI Ultra Access add-ons [1].
Where does translation not work?
Google lists several cases where speech translation is not available [1]:
- live streams;
- meeting recordings;
- client-side encrypted meetings;
- dial-in calls and third-party interoperability calls;
- meetings where adaptive audio is in use.
Availability also depends on regional and age restrictions [1].
How do privacy and admin settings work?
Each user decides whether their own speech is translated for others. They can change that choice at any time in Meet settings [1]. Google says it "does not store voice data or train models on your voice" [1].
For admins, speech translation is on by default for eligible editions when the Gemini for Meet setting is enabled [1]. Admins can switch it off or on per organisational unit or group. The setting is in the Admin console under Apps > Google Workspace > Google Meet > Gemini settings [1]. For users, translation is off at the start of every meeting and must be switched on during the call [1].
When will the feature appear?
Rapid Release domains get the full rollout from 22 October 2026. The feature may take 1–3 days to become visible [1]. Scheduled Release domains get a gradual rollout from 2 November 2026, which can take up to 15 days [1]. Mobile users need to update the Meet app. Google notes that visibility and synchronised captions may vary during the early rollout [1].
Who is affected?
- Facilitators of multilingual workshops. Translation is now a personal setting that starts off in every meeting [1].
- Distributed and international teams. Mixed-language meetings no longer need one agreed language pair [1].
- Hybrid teams using conference-room hardware. Rooms can hear translations but are not translated for others [1].
- Teachers and education teams. Access depends on the education AI add-on and on age restrictions [1].
- IT and workspace admins. The feature arrives switched on for eligible editions where the AI meeting setting is enabled [1].
- People who dial in or rely on recordings. They get no translation [1].
What should facilitators do now?
- Check the organiser's plan. In group meetings, translation for everyone depends on the organiser's plan, so the right person should send the invite. In 1:1 calls, either person's eligible plan is enough [1].
- Plan around the time limit. On base plans, translation stops being available after 90 minutes per meeting [1]. For long workshops, a scheduled break with a fresh meeting link may be the simplest fix.
- Brief participants at the start. Translation is off at the start of every meeting [1]. A short opening slide showing where the toolbar control sits will likely save time.
- Ask room speakers to join from a laptop. Speech from room hardware is not translated for others [1]. A second device with its own microphone may close that gap, but watch for audio echo.
- Do not rely on it for recorded or dial-in sessions. Recordings, live streams and phone callers do not get translation [1]. Shared written notes or a summary in each language may help those participants.
- Respect each person's choice. People decide whether their own speech is translated [1]. Mention this when you set ground rules.
What is not yet clear?
- The announcement gives the total of 63 languages but does not list them [1]. The Help Center may have the full list.
- Google mentions regional and age restrictions but does not say which regions or ages are affected [1].
- It is not stated what participants see when the 90-minute limit is reached, or whether translation can restart in the same meeting [1].
- Google gives no figures for accuracy or delay beyond "near real-time".
- It is not spelled out whose plan sets the time limit when the organiser and a participant hold different eligible plans.