Back to homeMac translation guide

Live screen translator for Mac: subtitles, apps, and changing text

A live screen translator watches one visual region and refreshes the translation when the words in that region change.

Platform
macOS 15+
Input
Pinned screen region
Source
Visible text, not audio

A screenshot translator solves one frame. That works for a static image or error dialog, but subtitles, shared screens, games, remote desktops, and foreign application panels keep moving. Capturing the same area again soon becomes the task.

TranslateWindow lets you pin one region and leave a small translation window beside it. The source stays visible, the result updates with the visual text, and you can keep watching or working.

What live screen translation does

Live mode repeatedly reads the pixels inside a region you choose. When the visible words change, TranslateWindow recognizes the new text and refreshes the translation. It does not translate the whole display or decide what to watch on its own.

Keeping the region small matters. A subtitle line, meeting caption area, chat panel, or software dialog is easier to recognize than an entire desktop full of unrelated text.

Where it is useful

Screen contentWhat to pinWhy live mode helps
Video or course subtitlesThe subtitle lineNew captions arrive in the same place
Meeting or presentationVisible captions or the shared-screen areaThe speaker or slide keeps changing
Game or foreign softwareDialog, menu, or status panelText appears inside an interface you cannot select
Remote desktopThe remote app panelmacOS cannot access the remote text layer
Support chat or dashboardThe updating message or status regionNew information appears without a page change

Visible text and spoken audio are different

TranslateWindow live mode translates text that is already visible on screen. It does not listen to system audio or create captions from speech. In a meeting, it can follow captions or words on a shared screen, but spoken audio without visible text needs a speech translation tool.

This boundary is useful when comparing live translation apps. Some products combine screen OCR, hover translation, overlays, and speech recognition. TranslateWindow keeps this mode focused on the changing visual region and leaves the original content visible.

A small result window instead of a full-screen overlay

TranslateWindow places the result in its own compact window beside the source. It does not replace every word inside the captured image. That keeps the original layout, labels, and visual context available for comparison.

The window can stay out of the way while a course, meeting, or application continues. When you are done, dismiss it and return to the same content rather than cleaning up a translated full-screen layer.

Use a scene for recurring terminology

A software demo, customer meeting, and university course may use very different language. Translation scenes can keep preferred terms, tone, and post-processing instructions with the task instead of using one generic setting for every screen.

If a result later needs to become an email, note, or reply, the same scene can guide text polishing. The live region handles what is changing now, while the scene keeps repeated wording consistent.

Set up a stable live region

  • Open the TranslateWindow command panel and choose live-window translation.
  • Drag over the smallest fixed area that contains the text you need.
  • Move the result beside the source so it does not cover the next line or control.
  • Choose the local model, cloud service, or scene that fits the content.
  • If the text is static, stop live mode and use one screenshot instead.

Privacy depends on the active model

A live region may contain customer messages, internal dashboards, and account details. With a local model, recognized text can stay on the Mac. If you choose a cloud service or your own API key, the required text is sent to that provider under its terms.

Keep the region tight and check the active model before monitoring sensitive content. TranslateWindow does not use screenshots, OCR text, source text, or translations as analytics content.

When another mode is better

Use selection translation for a paragraph you can highlight. Use screenshot OCR for a static menu, image, or error dialog. Use an audio translation product when the information is spoken but no captions are visible.

Live OCR also cannot recover text that is too small, blurred, or hidden. Increase the source size, improve contrast, or capture a smaller region before changing translation models.

Questions before you try it

Does it translate speech? No. Live mode reads text that is already visible. If a video or meeting has captions, it can follow those captions.

Can it translate a whole app window? You choose a fixed screen region. A smaller region usually gives cleaner OCR and keeps unrelated text out of the result.

Can it work offline? With a downloaded local model, the translation step can run on the Mac. Cloud services send the required text to the provider you choose.

Try live translation on one part of your screen

Pin a subtitle line or foreign app panel. TranslateWindow keeps the result beside it and refreshes when the visible text changes.

Download for macOSScreenshot translation guide

Sources and verification

Product facts were checked against these first-party sources. Links open the current official documentation.