So first of all, you should have a MKV file, with a VobSub subtitle file; or a combinaison of idx/sub files.

Extracting idx/sub files

When running $ mkvmerge -i mkvfile.mkv, you should see Track ID x: subtitles (VobSub) (where x is a number).

Do not forget to install the mkvtoolnix package, which contains mkvmerge.

Extract this vobsub file from your mkv by using my script from here.

1. chmod +x sub_mkv.sh
2. ./sub_mkv.sh mkvfile.mkv. When it asks “Subtitle Track ID:”, write the number x see above.
3. Normally you’ll now have two files: a .idx and a .sub (and optionnaly a .ifo)

## OCR the two files

### By using ffmes

1. Install ogmrip, tesseract-data-eng and tesseract-data-fre packages
2. Clone https://github.com/Jocker666z/ffmes
3. Put your idx/sub files and move them alone in a folder
4. Launch ffmes: bash ffmes.sh [FOLDER OF IDX/SUB FILE]/
5. Choose option 18, and the language
6. Done, you have the srt file!

### Old version: by using vobsub2srt

1. Then install vobsub2srt and a tesseract lang package (see all the lang package available here)
   • install tesseract-data-eng and tesseract-data-fre
   • vobsub2srt converts subtitles in VobSub (.idx / .sub) format into subtitles in .srt format.
   • execute it by using vobsub2srt mkvfile where mkvfile is the filename of the subtitle files WITHOUT the extension (.idx / .sub).
   • vobsub2srt writes the subtitles to a file called mkvfile.srt
   • It’s finished.

If a subtitle file contains more than one language you can use the --lang parameter to set the correct language. More info here
If you want to dump the subtitles as images (e.g. to check for correct ocr) you can use the –dump-images flag.