Live translation in Google Meet

Google Meet does voice-to-voice translation well and preserves the speaker's voice, but supports one language pair per meeting, all English-anchored, with a 90-minute cap and no translation in recordings or live streams. Two languages with English on one side is the case it fits.

Google Meet’s speech translation is good at the narrow thing it does. The limits are sharp rather than gradual, which makes it easy to evaluate: you are either inside them or you are not.

What it does

Voice-to-voice translation with the speaker’s voice preserved, powered by Gemini, included in Business and Enterprise editions with Gemini.

It covers 5 language pairs, all anchored to English.

The limit that decides most cases

One language pair per meeting.

A call with Italian, Spanish and German participants is not possible on Meet. Not expensive — impossible. This is architectural, not commercial, so no plan upgrade changes it.

Since every supported pair is English-anchored, two non-English languages cannot be paired with each other either. A Spanish speaker and a German speaker cannot use Meet’s translation to talk to one another.

The other three limits

90-minute session cap. Longer meetings, training sessions and workshops fall outside it.

No translation in recordings or live streams. The translation exists during the live session and nowhere afterwards. For a distributed team where most people watch later, this removes most of the value.

Five pairs. Check the current list against your actual languages rather than assuming coverage.

When Meet is the right answer

  • Two languages, one of them English
  • One of the supported pairs
  • Under 90 minutes
  • The live session is what matters, not the recording
  • Everyone on a Business or Enterprise edition with Gemini

That is a genuinely common situation — a bilingual team, a recurring call with one international colleague, a regular customer check-in. If it describes you, the feature is already in your licence and works well. Use it.

When it is not

  • Three or more languages
  • Two non-English languages
  • Over 90 minutes
  • Recording or live stream needs the translation
  • External participants outside your Workspace

How it compares on this specific point

Both alternatives trade one limit for another:

  • Teams Interpreter supports 10 languages rather than 5 pairs, but requires Microsoft 365 Copilot rather than Teams Premium, excludes external guests, and does not work in ad hoc meetings, webinars or town halls. Its recordings also drop the interpretation audio, by Microsoft’s own documentation.
  • Zoom covers 34 languages plus 3 target-only, but produces captions only — no translated audio.

There is no native option that handles three languages with translated audio in one meeting. That is the gap worth knowing about before you evaluate anything else.

Following a Meet call in your own language

Everything above is about what the organiser can turn on. There is a separate question: what can you do about a Meet call you were simply invited to, on someone else’s Workspace, where the pair is not English-anchored?

VoxTranslate for Chrome translates the audio of the Meet tab in your browser, showing subtitles over the call and optionally speaking the translation. It needs the web client — the desktop app has no browser tab to listen to — and it is one-directional: it translates what you hear, and the other participants get nothing back.

That is the honest boundary. It solves your comprehension, not the room’s. For a meeting where three people each need to be understood in their own language, the answer is still a tool that runs the call, not one that listens to it.

Practical notes

  • Brief speakers to pause at clause boundaries. Meet’s translation, like all of them, needs a phrase to work with.
  • Put numbers and names in the chat. Highest consequence, cheapest fix.
  • Test with real spontaneous speech, not a scripted run-through.
  • Decide what happens to the recording before the meeting, since the translation will not be in it.