Translated captions for Microsoft Teams on Mac

Check Teams’ own caption options first

Ordinary Teams live captions show speech in the meeting’s spoken language; translated captions add a different reading language. Teams has built-in captions on Mac, so check the meeting’s caption controls before choosing another tool. See Microsoft’s live captions instructions.

Microsoft’s current licensing guidance for translated meeting captions distinguishes the organizer from an individual attendee:

• Teams Premium or Microsoft 365 Copilot includes live translated captions.
• An eligible organizer’s license covers attendees: participants can use translated captions without each having their own eligible license.
• An eligible attendee’s own license covers that attendee: if eligibility comes from individual licenses instead, only those licensed participants get the Translate to option.

Source: Microsoft Support: Multilingual speech recognition in Microsoft Teams, checked October 23, 2026. Ask your organizer or IT administrator which options are available in your meeting.

If you do not have access to native translated captions, Sottola offers a separate way to read the audio on your Mac. Use it only where your organization and meeting participants permit recording or transcription; it is not a way to override meeting or workplace policies.

What Sottola hears in a Teams meeting

Remote participants’ voices arrive through Teams and play as audio on your Mac. Sottola captures that system audio, recognizes the spoken words, and translates the resulting text. Its floating caption window can sit alongside the meeting while you follow both versions.

This is audio capture, not a Teams integration: Sottola does not join as a participant, read Teams captions, or import the participant list. It does not send translated speech or captions back into Teams. Other attendees do not automatically receive your translations.

Your own microphone is not captured by this workflow. Do not expect a complete record of both sides of the call: your voice is absent unless it is separately present in the Mac’s system audio. Other apps’ playing audio can also enter the capture, so pause unrelated videos and music before recording.

Prepare before your first international meeting

You need an Apple Silicon Mac, M1 or newer, with macOS 15 or later. Download Sottola, drag it into Applications, and open it. Look for the waveform icon in the menu bar; Sottola has no Dock icon. Reopen its welcome window with Quick Start Guide in that menu.

1. Choose speech, languages, and translation; then prepare. In the guide, select your speech engine and spoken language. Choose Translate in the waveform menu, then open Settings → Model. Set From to the language you will hear, To to the language you want to read, and Method to a translation method using the options below. Return to the guide and select Prepare Selected Models; wait for Ready. For Apple Translation, also select Prepare Translation Languages… in Settings → Model and complete macOS’s prompts. Preparing models does not start your trial.
2. Grant system-audio access. Choose Request Audio Access, then enable Sottola in System Settings → Privacy & Security → Screen & System Audio Recording. If the installed app is missing, add it with the + button. This workflow does not require microphone permission.
3. Quit Sottola completely and reopen it. Use Quit Sottola, then open Sottola again from Applications. Closing the guide or caption window does not quit the app. The fresh process applies the permission change; continue when the guide reports access is ready.
4. Explicitly start your trial. Choose Start Seven-Day Trial in the guide, or activate your existing Sottola license. No payment card is needed for the trial. Trial activation requires internet and does not start recording.
5. Hear a remote speaker, then start captions. Join your permitted Teams meeting on this Mac and check that you can hear its audio through your selected speakers or headphones. In Sottola choose Start Recording, then Show Overlay. Look for the first original caption and then its translation. Test a short exchange before relying on this arrangement for a whole meeting.

For a low-pressure setup check before the call, play a short spoken sample on your Mac first. The first-recording guide walks through the same permission and trial sequence.

Choose the languages you hear and read

For an English-speaking meeting that you want to follow in French, set From: English and To: French. These are Sottola’s settings; changing them does not change the Teams meeting’s language. Speech recognition and text translation are separate choices, and both must support your pair.

• Nemotron speech is the default local route on macOS 15+. Choose the actual spoken language when you know it. Auto is available, but text translation still needs a supported detected language.
• Apple On-Device speech is an optional route on macOS 26+. Select an explicitly supported language and region, then prepare its Apple-managed speech assets. It does not support Auto.
• Local Translation with TranslateGemma works on Apple Silicon and macOS 15+ after downloading the selected model: about 2.2 GB for 4B or 6.7 GB for 12B. Runtime memory needs exceed download size.
• Apple Translation separately needs macOS 26+, a supported language pair, and installed translation packs. Preparing Apple speech does not prepare these packs.

Use the live translated captions guide and language lists to check your choices. Selecting the same source and target turns translation off. In a meeting that switches languages, check recognition against what is being said; one selected source language is not a promise of automatic multilingual interpretation.

With local speech and translation selected, Sottola processes the captured audio and text on your Mac after preparation. Teams still carries the meeting online. Sottola’s optional cloud engines are separate, explicit choices; selecting them changes where the relevant audio or text is processed.

Follow captions, then stop and save

Read the original line alongside the translation. Recognition can revise a phrase as the speaker finishes, and translation may arrive later. An Updating indicator means a translation is catching up with changed source text. Check names, dates, numbers, and important decisions against the original rather than treating a draft as final.

When you finish, choose Stop Recording in Sottola and let pending recognition and translation finish. With transcript auto-save enabled in Settings → Display, the completed conversation is saved locally and available in History. Open History to review, copy, or export it; this is Sottola’s transcript, not a Teams meeting transcript.

If auto-save is off, save or export the current conversation before quitting. Saved transcripts remain available to read and export after the trial ends. The recording only includes the audio Sottola captured while it was running.

If no caption or translation appears

• No original text: check that a remote participant is actually speaking and you can hear Teams on this Mac. Silence between speakers is normal. If you just granted access, stop, fully quit Sottola, and reopen it.
• Original text but no translation: confirm Translate mode, different From and To languages, and readiness of the selected text translator. Apple Translation needs its own installed pair; Apple speech readiness is not enough.
• Inaccurate words: check the spoken language and listen for overlapping voices or unclear audio. Translation inherits recognition mistakes. Sottola does not obtain participants’ names from Teams; optional Nemotron speaker labels are anonymous, and Apple speech has no speaker labels in Sottola.

Stop recording before changing a speech or translation route, prepare any newly selected assets, then try a short passage again. Test your own language pair and meeting audio during the trial; this guide does not promise a measured delay or accuracy level.

See a separate English-to-French sample

The public Sottola demonstration pairs a fictional narrated English planning conversation with French translation. You can listen to its source audio and inspect the unedited transcript and notes. It demonstrates the caption workflow with a sample recording, not a Microsoft Teams call.