Blog

The Meeting Can Translate. The Switch Belongs to Whoever Booked It.

November 9, 2026  ·  Tablingo

Zoom, Meet and Teams can all translate a meeting. Whether they will in the one you are sitting in was decided before you joined — by a licence, a checkbox and a language list.

You are in a call and the language is not yours. Maybe you catch seven words in ten, which sounds like enough until someone asks a question and the answer turns on a distinction you missed.

At some point a colleague will offer the reassuring fact: the platform translates now. It is true. Zoom, Google Meet and Microsoft Teams all ship a way to turn speech into another language, and two of them have been doing it for years.

But “the platform can” is a sentence about a product. It says nothing about the meeting you are in. For that you need a different question, and it is not the one in the help centre: who holds the switch, and when did they hold it?

Three platforms, three owners

Here is the landscape in one table. Read the third column, not the second.

PlatformWhat it offersWho turns it on
ZoomHuman interpretersThe host, before the meeting
Google MeetTranslated captionsYou, if your edition has it
Microsoft TeamsCaptions, AI interpreterThe organiser’s licence

The second column is what every comparison article prints, and it is why they all read the same. The third column is the one that decides whether you see anything.

Zoom: a scheduled meeting with an interpreter roster

Zoom’s answer is not machine translation. It is human interpretation, delivered as separate audio channels: an interpreter speaks into a channel, and you pick the channel.

Which is a good design, and completely gated:

The client matters too. Interpretation cannot be started or managed from the mobile app or the web client; participants who joined that way can only select a channel and read the interpreted text.

So count what had to be true before you clicked the link: somebody’s paid licence, a meeting that was scheduled rather than started, a particular ID format, and an interpreter assigned to a language. You contributed none of it.

Google Meet: the one place the switch is yours

Meet is the exception, and it deserves to be stated plainly. In a meeting you can open More options, then Settings, then Captions, set the language of the meeting, turn on translated captions and pick the language you want to read. No host action required.

Two conditions remain, and neither is yours:

Still, this is the honest counterexample: on Meet, if your account is covered, you can solve your own problem without asking anyone. On the other two you cannot. That difference is the whole article.

Teams: a licence that cascades, and a list you cannot extend

Teams has three separate mechanisms, licensed three different ways, which is why its help pages are so hard to read.

Then the detail that makes the mechanism visible. In a town hall or a live event, the organiser preselects which languages will be offered: six of them, or ten with a Premium licence, out of more than fifty available. Attendees can only choose from that list. Somebody wrote the menu a day earlier, and your language may not be on it.

One more property worth knowing, because it inverts the usual complaint about transcripts: translated captions and translated transcripts in Teams exist during the meeting and are not accessible after it ends. The translated layer is not saved. The recording keeps the original audio.

That turns out to be the general rule rather than a Teams quirk — what a recording keeps is a second pass, not the layer anyone watched. It is the subject of its own article, including the dropdown in Zoom that looks like translation and is not.

The same three gates, twice

Set Zoom and Teams side by side and the pattern stops looking like a coincidence.

What it takesZoomTeams
A licenceThe host’sThe organiser’s
Set up in advanceWhen schedulingWhen scheduling
A language listInterpreters assignedPicked by the organiser

Three gates, always in the same order: a licence someone bought, a decision someone made while scheduling, and a list of languages someone wrote. Not one of them is a per-viewer setting. All three are properties of the meeting — and a meeting belongs to the person who booked it.

This is not an oversight

It would be easy, and wrong, to read the above as three companies failing to ship a feature. They shipped it. The design is deliberate and it is aimed at a real customer: the all-hands with interpreters booked, the board call between two offices, the webinar with a paying audience in four countries. Those meetings know in advance that they are multilingual, and for them a roster of interpreters is the correct answer rather than a limitation.

What the design does not cover is the other kind of multilingual meeting, which is far more common: the one nobody planned for. A supplier calls. A candidate interviews. A student joins a lecture. A reporter sits in on a briefing. The call is in a language that one person in it does not read, and nobody arranged anything, because until that person joined the meeting was monolingual.

In that room the feature exists, is switched off, and cannot be switched on from where you are sitting. Not because the vendor is stingy — because the switch was never on your side of the screen.

If you are the one who booked it

Then all of this belongs to you, and the advice is short:

  1. Zoom. Before the meeting, open it in the web portal, enable language interpretation, add each interpreter’s email and target language, and keep the meeting ID on automatic.
  2. Teams. Check that the organiser holds a Premium or Copilot licence, and turn on multilingual speech recognition for the meeting if people will switch languages mid-sentence.
  3. Meet. Confirm the edition includes translated captions, and leave the voice-translation admin setting enabled if you want the spoken version.

If you control the invitation, use what the platform already gives you. It is better than anything bolted on, and you have already paid for it.

What is left for everyone else

Which leaves the position most people are actually in: you did not book it, you cannot buy the licence, and the language list was fixed yesterday. You can ask the host, and asking works sometimes. But asking a room of twelve, mid-meeting, to reconfigure the call for you is a different social act from clicking a switch, and most people would rather struggle through than do it.

This is where our extension sits, and why it is shaped the way it is. It reads the audio of the tab you already have open, transcribes it as it plays, and puts the original line above the translation on the page. There is no licence to check, no organiser to convince, no list to be added to. Nobody has to have planned for you in advance.

Two boundaries, stated plainly. It works inside a browser tab, so a meeting running in the desktop client is out of reach — if the invite has a “join in your browser” link, that is the way in, and the captions work there. And it is machine translation: good enough to follow a call, not a substitute for a professional interpreter when the words are contractual.

Two checks

  1. In your next meeting, open the caption settings and look for “Translate to”, or the interpretation globe in the toolbar. If it is missing, you have found the switch — and it is not yours.
  2. Ask who booked the meeting. If the answer is not you, the language layer was decided before you arrived, by a licence and a checkbox you never saw.

Both checks take ten seconds and require no trust in any vendor, including us.

What you actually wanted

Strip away the products and the request is small: the words, while they are being said, in a language I read, in a meeting I do not control. Three of the largest meeting platforms built exactly that, and then attached all three switches to the same object — the invitation.

If you own the invitation, use what you already pay for. If you do not, the layer has to come from your side of the screen, because nobody is going to move it to yours.

If you want to try ours, it's called Tablingo — a browser extension that reads the tab you are already in and puts the original line above the translation. The general meeting case, including the platform-by-platform setup, is here, and why a caption track and captured audio behave differently is here.