Live translation in Zoom, Google Meet and Microsoft Teams: what each offers

Captions or speech: two kinds of live translation

Translated captions show text of what others say in a language you read. Speech translation goes a step further and speaks a translated voice. Captions are the older and wider feature. Speech translation is newer, covers fewer languages, and usually needs a more expensive licence.

Zoom

Translated captions. Included in Workplace Business Plus, Premier, Enterprise Essentials and Enterprise Plus, or an add-on to any paid licence. Zoom's blog from Feb 2024 gave 35 languages and $5 a user a month for the add-on. We could not confirm the current price or language count on a help page, so check the vendor page. The host's caption settings decide which languages are available, and translation is not available when manual captioning is on.

Voice translator. Speech to speech, in 9 languages: English, Chinese, French, Japanese, Spanish, German, Italian, Arabic and Portuguese. It needs a Basic, Pro, Business or Enterprise account plus the Live Translation add-on or ZoomMate, the desktop app 7.0.0 or later, and AI Credits apply. Each person controls their own settings, and settings can only be configured during a meeting. Whether a host or admin must enable it, and whether it is explicitly two-way, are not stated on the page we read. Check the vendor page.

Google Meet

Translated captions. 69 languages listed, on Business Standard, Business Plus, Enterprise Standard, Enterprise Plus and Google AI Pro for Education. Some languages are in beta, so quality may vary.

Speech translation (Gemini). 63 languages listed, delivered 'in a voice like yours' with captions. It needs Google AI Pro or Ultra, or an eligible Workspace edition. For scheduled meetings the host needs the eligible plan, and others must opt in. Per-meeting limits apply: 90 minutes on most plans and 8 hours on AI Ultra and AI Expanded Access. Google says translations are delayed by a few seconds, and it is not available in live streams or recordings.

Microsoft Teams

Live translated captions. The 'Translate to' list has 31 languages. It needs Teams Premium or Microsoft 365 Copilot. Each participant turns it on, the organiser's licence covers everyone, and admins can disable it. Captions and transcripts exist only during the meeting. Microsoft notes that accuracy drops with distance from the microphone, background noise and overlapping speech.

Interpreter agent. Speech to speech, using a pipeline of speech recognition, machine translation and text to speech. It needs a Microsoft 365 Copilot licence. Microsoft's Learn page lists 9 languages, while the support page lists 10, so the figures conflict. Check the vendor page. It is not supported in ad-hoc meetings, webinars, town halls, Teams Free or phone (PSTN) calls. Each user gets 20 hours a month with Copilot. Microsoft says accuracy is not guaranteed and the tool is not optimised for rapid exchanges, interruptions or overlapping dialogue.

Side by side

Limits all three state

• Accuracy is not guaranteed. All three say so. Names, technical terms and genders can come out wrong.
• Noise and overlap hurt. Poor audio, background noise, overlapping speakers and weak networks lower quality.
• Recordings do not carry the translation. Zoom captions are not on the recording, Teams records the original audio only, and Meet speech translation is not in recordings.
• Plans gate everything. Each feature needs a particular plan, licence or add-on, and often an admin setting.
• Some languages are beta. Google marks some caption languages as beta.

The gap: what do I say back?

Based on the help pages we read, none of the three describes suggesting what you should say back, drafting a reply or coaching tone. They translate what other people say, and Meet and Teams also translate your own speech for others. That is not proof the feature cannot exist. It is just not described on those pages.

That gap matters if you are the one working in your second language. Understanding a question and finding the right answer in the same few seconds are two different jobs. See also speaking confidently in meetings in a second language.

What Palmcard does differently

Palmcard is a desktop app for Mac and Windows. It listens to the call through your computer, so it works with Zoom, Meet, Teams, Webex and Slack huddles without joining as a bot. You choose the language you read best and the language spoken on the call, from 25 languages. It transcribes the call in that language and writes your next line in it, with the meaning shown in your own language. It does not replace the platform tools above. It adds the reply.

English uses the fastest transcriber, and the other languages use a multilingual one that is slightly slower. In our tests the line appears 0.5 to 1.1 seconds after the other person stops talking. It is a suggestion, so you stay responsible for what you say. See live translation for calls and Palmcard pricing.