Google Meet expands speech translation to 63 languages

Google announced on October 22, 2026, that speech translation in Google Meet now supports 63 languages. The update lets participants choose their own listening language while Meet detects the language a speaker uses. Availability still depends on the organizer’s plan, the rollout schedule and the kind of call. A team planning a multilingual meeting should check those conditions before relying on the feature.

What changed from the earlier version?

Meet’s April mobile rollout supported translation between English and Spanish, French, German, Portuguese or Italian. Only one language pair could be active in a meeting. Mobile access, therefore, is not the October news. The new announcement replaces that meeting-wide pair with automatic language detection and a separate listening choice for each participant. Google says a speaker can switch languages during a call and listeners can receive translations into their chosen languages. Its current help page lists the supported languages, including Arabic.

Those are Google’s product claims, not a measurement of accuracy or speed by this site. The announcement calls the audio translation near real time. Google Help says translated speech can lag by a few seconds and that captions may occasionally identify a language incorrectly or arrive late. For a high-stakes conversation, a spoken translation should be treated as an aid to communication, with important details confirmed in writing.

Can your next meeting use it?

Start with the organizer. Google’s eligibility guidance says a scheduled meeting needs a host on an eligible plan. In a direct one-to-one Meet call, at least one participant needs an eligible plan. Guests from another organization can use translation in an eligible meeting; they do not each need to buy the host’s plan. For Workspace, Google lists Business Standard and Plus, Enterprise Standard and Plus, Frontline Plus, Google AI Pro for Education, and the AI Expanded and AI Ultra Access add-ons. It also lists Google AI Pro and Ultra consumer plans. Check the current plan list rather than assuming every Workspace account qualifies.

Next, check when the new controls will appear. Google’s rollout notice says full rollout for Rapid Release domains begins October 22 and takes one to three days for feature visibility. Scheduled Release begins November 2 and can take up to 15 days. Google warns that synchronized captions and feature visibility may vary during the initial rollout. A missing toolbar control on October 23 does not, by itself, prove an account is ineligible. Mobile participants need an updated Meet app for the new capabilities.

Then confirm whether the meeting format is supported. Google excludes live streams, recordings, client-side encrypted meetings, dial-in and third-party interoperability calls, and calls using adaptive audio. Regional and age restrictions also apply. Google Help names under-18 users and unsupported devices among reasons translation may be unavailable. This makes an ordinary Meet meeting on supported web or mobile clients a safer choice for a first trial than a mixed dial-in or room-based event.

Finally, ask the people in the call to turn it on. Translation is off for individuals at the start of a meeting, even where an administrator permits it. Participants choose a language to hear and separately consent before their own speech is translated for others. According to Google’s use instructions, the control is under Captions & translation → Translate speech. Another participant will not hear a translation unless they enable the feature too. Workspace administrators can disable it for an organizational unit or group; the eligible Workspace setting is on by default when Gemini for Meet is enabled.

The limits that matter in a real call

Duration can be a deal breaker. Google allows up to 90 minutes per meeting on the eligible base Workspace editions and Google AI Pro. AI Expanded Access, AI Ultra Access and Google AI Ultra have an eight-hour limit. These are translation limits, not a promise that every meeting will stop at that point. If translated discussion is central to a long workshop, check the organizer’s tier and plan a fallback before the 90-minute mark.

Conference-room hardware has a separate constraint. Someone using Google Meet hardware can hear translated speech from a remote participant, but speech coming from that room’s hardware will not be translated for remote listeners. This limitation was already in Google’s April notice; the language expansion does not remove it. An illustrative case is a hybrid meeting where a remote colleague needs to understand a presenter speaking through the room microphone. The presenter should not assume that Meet will translate that room audio. A supported individual device or another human interpretation arrangement may be needed, subject to the organization’s meeting rules.

Translation also needs mutual consent and a little patience. Google says listeners only hear translated speech from a participant who has agreed to be translated. Its help page warns of grammatical errors, unintelligible words, unexpected accents and quality changes on a poor network. For decisions involving names, figures or commitments, repeat the key point and confirm it in meeting chat or a follow-up note. That is an editorial precaution based on Google’s stated limitations, not a claim from a test we ran.

What to try now, and what remains unknown

For a first use, the practical sequence is short: confirm the organizer’s eligible plan and release track; use a supported Meet client; update the mobile app if needed; ask participants to select their listening languages and opt in to translating their own speech; then run a brief rehearsal with the actual languages and devices planned for the meeting. If the control is absent, check rollout timing and the administrator setting before changing plans. If captions help a participant follow along, Google says they can be enabled alongside translated audio.

The announcement does not provide a comparative accuracy score for all 63 languages, a precise latency benchmark, or a guarantee that every eligible customer has the new interface today. We have not tested it. Readers evaluating translation tools more broadly can use our DeepL Voice guide for another product’s context, our AI subscription comparison for wider plan-selection questions, and our enterprise AI adoption guide for planning a rollout. For Meet’s current capabilities and access rules, Google’s linked announcement and help page remain the sources to check.