Embedded vs burned-in subtitles

Subtitle extraction depends on whether the captions are a real text track or part of the video image.

Quick answer

Embedded subtitles can often be extracted. Burned-in subtitles cannot be extracted as text without OCR.

Use Extract Subtitles from Video when the video file contains an embedded text subtitle stream.

Embedded subtitles are separate streams stored inside a video container.

Common examples:

MKV file
- video stream
- audio stream
- subtitle stream: subrip
- subtitle stream: ass

These can often be extracted into a separate .srt, .vtt, or .ass file.

Burned-in subtitles are pixels in the video itself. A player cannot turn them off because they are part of every frame.

They look like subtitles, but there is no subtitle file inside the video to extract.

Try these checks:

If the player has a subtitle menu and can turn captions off, they may be embedded.
If the subtitles are always visible and cannot be disabled, they are probably burned in.
If FFmpeg lists a subtitle stream, it is embedded.
If FFmpeg finds no subtitle stream, extraction will not produce text.

MKV supports subtitle streams, but a specific file may still have none.

OCR is a different workflow. It reads text from video frames and often needs manual correction.

Some embedded subtitle streams are image-based, such as PGS. They are embedded, but not plain text, so conversion to SRT may fail.