Translated captions for Google Meet on your Mac

Choose Meet captions or a separate Mac overlay

Google Meet has its own caption controls. For its built-in translated captions, open the meeting’s More options → Settings → Captions, choose the meeting language, enable translation, and select an output language. Availability depends on eligible editions and languages; check Google’s translated-caption requirements for your account.

This guide covers Sottola’s separate Mac app. Its captions sit above your other windows and can be saved as a transcript. They are not inserted into Meet’s caption panel, sent to other participants, or used to translate your voice for the call. Sottola is an independent product and is not affiliated with Google.

Before the call: check the Mac and audio

You need an Apple Silicon Mac with macOS 15 or later, Sottola, and meeting speech playing through that Mac. The system-audio path does not capture your microphone. It can caption remote voices you hear through Meet; your own spoken replies are absent unless they are separately played into system audio. Audio coming only from a phone or meeting-room device is also outside this Mac’s capture.

Before joining, use Meet’s speaker selector and Test speakers control in the preview to confirm the intended output. Google documents the speaker check and audio-device controls. The test sound checks playback, but it is not speech: use a spoken sentence for the caption test below. Pause unrelated videos or music so they do not mix with meeting speech.

Set this up before the meeting you need to follow. Initial model downloads and trial activation need internet access. With local speech and local text translation selected, Sottola processes the captured audio and text on your Mac; Google Meet still carries the call over its own service.

1. Choose languages and prepare models

Install Sottola in Applications and open its Quick Start Guide. If the guide is closed, find it in the waveform menu at the top of your screen. Choose your speech engine before preparing assets:

• Nemotron (Local Streaming) is the default and works on macOS 15+. Select the language being spoken. Automatic language selection is available, but an explicit source is useful when testing translation.
• Apple On-Device needs macOS 26+, a supported language and region from your Mac’s list, and prepared Apple speech assets. Select the actual spoken locale; Auto is unavailable.

Choose Translate in Sottola’s waveform menu, then open Settings → Model. Set From to the speaker’s language and To to the language you want to read. For example, an English-speaking meeting with French captions uses English as the source and French as the target. Check the speech and translation language information for your chosen provider; different providers do not support every same pair.

In the same tab, set Method to Local Translation using TranslateGemma for macOS 15+, or Apple Translation for macOS 26+ with a supported pair and installed packs. Speech recognition and text translation have separate assets. Apple speech being ready does not mean Apple Translation packs are ready.

Return to the guide and choose Prepare Selected Models. Wait for the selected speech engine and any TranslateGemma model to be ready. For Apple Translation, also select Prepare Translation Languages… in Settings → Model and complete macOS’s prompts. Preparing models does not start your trial. The translation-method comparison explains model downloads and language limitations.

2. Allow audio, then fully quit and reopen

1. Choose Request Audio Access in Sottola’s guide.
2. In macOS System Settings → Privacy & Security → Screen & System Audio Recording, enable Sottola. If it is missing, use + to add /Applications/Sottola.app.
3. Quit Sottola completely using Quit Sottola in the guide or Quit in its waveform menu. Closing the guide or overlay window is not enough.
4. Reopen Sottola from Applications and return to the guide. The newly granted permission must be checked in the reopened app before continuing; prepared models are reused.

3. Start the trial and test a spoken sentence

When models and audio access are ready, choose Start Seven-Day Trial in the guide, or use your existing active license. No card is required. This starts the trial clock; recording remains a separate action.

1. Join Meet with audio on this Mac. Confirm that you can hear the other person. Use a short test call before relying on captions for a meeting.
2. Start Recording in Sottola. Use the guide or waveform menu, then choose Show Overlay if you need to bring the floating captions into view. Position it where it does not cover Meet’s controls.
3. Ask the other person to say a short sentence. Use the source language you selected. Your own microphone test will not prove this capture path works.
4. Check original text, then translation. Confirm that the original matches the spoken sentence and the second line is in your target language. Recognition and translation can update as the phrase finishes; a translation can arrive after its original.

For a separate example before joining a call, play our public narrated English-to-French demonstration. It includes the source audio and completed output. That fictional narration demonstrates Sottola’s system-audio workflow; it is not a Google Meet test or evidence of Meet-specific accuracy or speed.

Check the overlay before screen sharing

Your Sottola window stays on your Mac unless you expose it through screen sharing or another recording. Meet offers tab, window, and entire-screen presentation; see Google’s presentation instructions. Choose the specific content you intend to present and inspect the sharing preview. A separate content tab or window also avoids presenting the meeting back to itself.

Do not assume the overlay is hidden. An entire-screen share may include captions, and capture behavior can vary with the browser, macOS, and selected sharing mode. Check from a recipient’s view in a test call. Hide the overlay before presenting if you have not verified the result or if the captions should not be visible. Sottola does not add a private-caption guarantee to Meet’s sharing controls.

Stop and keep the completed transcript

Choose Stop Recording in Sottola when you finish. Leaving Meet alone does not stop Sottola’s system-audio capture. Wait for pending speech and translation to finish before reviewing the result.

To keep a file, enable Auto-save transcripts on stop in Settings → Display before recording. Open History to review, copy, or export saved conversations. Check names, numbers, and decisions against the original wording before sharing a transcript. Captions are generated text and can contain mistakes.

If captions or translation are missing

The optional setup-count setting is off by default and is not required for captions or trial access. See Sottola’s privacy details before enabling it.