Skip to content
TrackHeros

Answer

Does automatic language detection work on Microsoft Teams?

For people who take client calls on Microsoft Teams in more than one language: whether TrackHeros detects the meeting language on Teams as it does on Google Meet, what the Teams captions can and cannot do, what happens after the call, and which language the notes are written in for a meeting held in another language.

By · Updated 27 October 2026 · 3 min read

Automatic language detection in TrackHeros does not depend on the meeting platform. This answer is for consultants, developers and anyone else who takes client calls on Microsoft Teams in Spanish, French, German or any of the close to 100 languages TrackHeros recognises, sometimes two in one call. The full explanation is on automatic language detection, and how the bot behaves in a Teams call is on the Microsoft Teams notetaker.

What can the Teams captions do, and what can't they?

They give the real speaker names, and they decode one language at a time. A Teams meeting's captions are set to one spoken language, so speech in any other language is forced through a decoder that expects the wrong one. The result is text in the caption language that has little to do with what was said.

That is why the captions are treated as a scaffold, not the transcript. They say who spoke and when. Whether they also say what is decided meeting by meeting. Google Meet captions work the same way, which is explained in why meeting captions pick one language.

What happens after a Teams meeting ends?

TrackHeros works out the meeting's language in stages, cheapest first, and the stages are the same on Teams as on Google Meet:

  1. A recurring meeting's memory of the language it has been held in.
  2. A check of whether the captions read as coherent text.
  3. A short audio sample, about two minutes, asked only which languages are spoken.
  4. A probe, and then a full audio pass over the recording.

Any stage can send the meeting to the audio pass; none can keep it on the captions alone without independent confirmation. English-only meetings use Google's Gemini model on Vertex AI. Meetings in another language, or in more than one, are transcribed by a Whisper model, run by DeepInfra, Together AI or Groq through OpenRouter, that detects the spoken language itself. The speaker names still come from the Teams captions, and a line that cannot be attributed is labelled "Unknown". More in how a notetaker can detect the language on its own.

Is anything different from Google Meet?

Not in the language decision. The differences between the two platforms are about getting in: on Teams the organiser's lobby settings decide whether the bot is admitted automatically or waits to be let in. Once the meeting has been recorded, the platform no longer matters, because the second pass listens to the recording rather than to Teams. The Google Meet side is on the Google Meet notetaker.

What do the notes look like for a Teams meeting in another language?

The notes are in the language you choose; the transcript is in English. The transcript is rendered into English line by line, each line keeping its speaker and timestamp, and a line that fails to translate keeps its original text rather than disappearing. The summary, decisions, action items and minutes are then written from that transcript in English, or in Arabic, French, Spanish, German, Italian, Portuguese, Ukrainian or Indonesian — set for your account, a client, a recurring meeting or a single call. A Teams call held mostly in German can get German notes on its own, with "same as the meeting".

The transcript page says one thing about language: "Translated to English from a conversation held in German", or whichever languages were heard. Every language TrackHeros recognises, and the nine the notes can be written in, are on the languages page.

Related

Early access

Get in early.

We’re onboarding a small group at a time so we can get it right. Join the list and we’ll reach out.

Get early accessEarly access · Google Meet and Microsoft Teams