Skip to content

How to Translate Video Subtitles in Real Time on Mac

/ 7 min read
macos tutorial translation productivity video

You found a lecture that explains exactly the thing you have been stuck on for a week. The catch: the whole talk is in Japanese, and the subtitles are burned right into the video. You cannot download an .srt file, you cannot select the text, and pausing every few seconds to type a sentence into Google Translate turns a 40 minute video into a two hour slog.

Hardcoded subtitles are one of the most annoying translation problems on a Mac. The text is right there on screen, readable and sharp, but it lives inside the pixels of the video frame. Standard tools assume you can copy the words. When you cannot, you need a different approach. Here is how to translate video subtitles line by line, in real time, without leaving your seat every ten seconds.

Why Video Subtitles Are So Hard to Translate

Not all subtitles are equal. Some players load subtitles from a separate text track, which you can swap or auto-translate. Hardcoded subtitles are different. They are rendered permanently into the picture, the same way a caption is baked into a meme. Common sources include foreign films from small distributors, online course platforms, conference recordings, fan-subbed clips, and anything screen-recorded from another device.

Because the text is part of the image, the usual tricks fail:

  • You cannot right-click and choose Translate, because there is nothing selectable.
  • You cannot download the subtitle file, because there is no separate file.
  • Player-based auto-translation only works when a real text track exists.

That leaves you with OCR, optical character recognition, which reads the characters straight from the frame. If you want the background on how screen OCR turns pixels into readable text, this OCR translation guide for Mac walks through the full pipeline.

The Slow Way: Screenshot and Paste

Before the good method, it helps to see the painful one. The manual loop is pause the video, screenshot with Command Shift 4, open a browser tab, upload to an online translator, wait, read, switch back, unpause, repeat for the next line.

For one caption that is fine. For a 30 minute talk with a new subtitle every few seconds it is unworkable. You spend more time managing screenshots than watching, and every frame you upload leaves your machine, which is a real concern for a paid course or internal training material.

The Fast Way: Live Mode OCR Translation

The workflow that actually keeps up with a video is one that reads each subtitle as it appears and translates on the fly, without a manual screenshot each time. This is exactly what Live mode in TranslatePeek is built for.

Here is the shape of it. You point the tool at the region of the video where subtitles show up. As new lines appear, they keep getting translated, line after line, so you read the translation in the same spot while the video plays. There is no pausing, no browser tab, no upload step. Press Command Shift L to start Live mode, aim it at the subtitle area once, and let it run.

A few things make this comfortable in practice:

  • It targets a region, not the whole screen. You tell it where subtitles live, so it is not distracted by menus, timelines, or on-screen graphics.
  • It keeps translating as lines change. You are not firing a shortcut per subtitle. The translation refreshes as the text on screen changes.
  • The floating panel can read the translation aloud and copy it, which is handy when you want to keep your eyes on the visuals instead of the text.

TranslatePeek covers 19 languages in both directions, including Japanese, Korean, Chinese Simplified and Traditional, Spanish, French, German, Russian, Thai, Vietnamese, and more. So a Korean documentary, a French cooking course, or a German engineering lecture all work the same way.

Setting Up for a Watch Session

A little setup makes a long video much smoother.

Get the layout right first. Play the video in a window rather than true full screen, so the floating translation panel has somewhere to sit. Borderless or windowed mode both work. Position the video so the subtitles sit in a stable spot near the bottom.

Aim once, then relax. Start Live mode with Command Shift L and mark the subtitle region a single time. As long as the video does not move, you will not need to re-aim.

Pick your direction. Set the source and target language before you start, for example Japanese to English. Automatic detection handles most cases, but locking the source language helps when a video mixes on-screen English UI with foreign dialogue.

If a particular line comes out garbled, it is usually a contrast or size issue rather than a language issue. The next section covers that.

When Subtitles Come Out Wrong

OCR is strong but not psychic. A few situations trip it up, and each has an easy fix.

Low contrast subtitles. White text on a bright, busy background is the classic failure. Many players let you nudge playback brightness, and pausing on a darker frame often helps the text stand out.

Stylized or decorative fonts. Fan subs and stylized films sometimes use fonts that OCR struggles with. Capturing a slightly larger region gives the recognizer more to work with.

Fast subtitles. If lines change very quickly, pause for a beat on the important ones. Even in real-time reading, a short pause on a dense sentence is worth it.

Because everything runs with Apple’s on-device Vision and Translation frameworks, none of this depends on your connection. A weak hotel wifi or an offline flight makes no difference, and no frame of your video ever leaves the Mac.

Beyond Video: The Same Trick Everywhere

Once real-time OCR translation is set up, you start reaching for it constantly. The same approach handles on-screen game text on Mac, where dialogue and menus are rendered as graphics, and it handles a single translate a screenshot to English job when you just need one frame done rather than a whole video. Video subtitles are simply the case where reading line after line, without breaking flow, matters most.

FAQ

Can I translate hardcoded subtitles without pausing the video?

Yes. With a Live mode OCR tool like TranslatePeek, you aim at the subtitle region once with Command Shift L and the translation refreshes as new lines appear, so you can keep watching at normal speed.

Does this work if the video has no separate subtitle file?

That is exactly what it is for. OCR reads the text directly from the video frame, so it does not matter whether a downloadable subtitle track exists. Burned-in captions work the same as any other on-screen text.

Is real-time subtitle translation private?

With an offline tool it is. TranslatePeek uses Apple’s on-device frameworks, so nothing about the video is uploaded. That matters for paid courses, internal training, and anything you would rather not send to a server.

Which languages are supported?

TranslatePeek handles 19 languages in both directions, including Japanese, Korean, Chinese, Spanish, French, German, Russian, Thai, and Vietnamese, which covers the large majority of foreign film and course content.

What are the requirements?

TranslatePeek needs macOS 15 or later and runs on Apple Silicon or Intel Macs with a Neural Engine. It is a one-time purchase of $6.99 with no subscription.