Extract subtitles from .mkv files on Mac OS X

subtitle-extract.md

Extract Subtitles From .mkv

These instructions shall work on Mac OS X and Linux.

Install mkvtoolnix (Mac OS X)

Installation via Homebrew:

brew install mkvtoolnix

List content of the .mkv file

mkvmerge -i myFile.mkv

Example of the file content

File 'myFile.mkv': container: Matroska
Track ID 1: video (V_MPEG4/ISO/AVC)
Track ID 2: audio (A_DTS)
Track ID 3: subtitles (S_TEXT/UTF8)
Track ID 4: subtitles (S_TEXT/UTF8)
Chapters: 22 entries

Track ID 3 and 4 are subtitles.

Extract subtitles tracks into separate files

mkvextract tracks myFile.mkv 3:myFile1.srt 4:myFile2.srt

Enjoy!

---

Extract Multiple Files

In case you need this in batch for all files in a directory (works only if all files have the same subtitle Track IDs).

Make sure you use correct file extension (*.mkv) and track-ids (3)

Example:

for file in *.mkv; do
  sub=$(echo $file | sed 's/\.mkv$/.srt/'); 
  mkvextract tracks "${file}" 3:"${sub}"; 
done

Thinks.Help me a lot.

Written really well, even for newbies.

thank you, this helped a lot.

Amazing, thank you! I have almost no idea what I'm doing, but I managed to figure it out (like I barely know how to use the commands on the terminal app on my Mac). :)

Alternate:

0. Install ffmpeg (for example, through brew install ffmpeg)
1. ffmpeg -i video.mkv
2. Look for Stream #<stream number>(eng): Subtitle: subrip where <stream number> is the stream number of the subtitle track. Eg: 0:2
3. ffmpeg -i video.mkv -map <stream number> subs.srt (example: ffmpeg -i video.mkv -map 0:2 subs.srt)

I'm trying this, but I keep getting the error:

-bash: syntax error near unexpected token `('

I guess it's because the file name contains parentheses ()? How can I solve this? Any help appreciated.

Hi guys, I forgot that even exists. Just better formatted @bmaeser notes :)
Anyway, I am quite happy for your appreciation.
Feel free to suggest any improvements.

I added useful notes "Extract Multiple Files" from @farzadshafiee

Thanks a lot! That was really helpful!

Hi I keep getting the message that Error: The file 'Episode_1.mkv' could not be opened for reading: open file error.

@pavelbinar

I did exactly as suggested. It extracted a file. The file ended in .srt because that's the filename I specified, but it's a 25MB binary file, not a plain text .srt file. I suspect it's a .sub file format. Here are the first couple of lines:

PG�9ÅÆ��������Ä�8���Ä��������˛�mPG�9ÅÆ������
���˛�m�á�îPG�9ÅÆ������B����ÄÄÄ��ÄÄ@��ÄÄ`��ÄÄø��ÄĘ��ÄÄ���ÄÄü��ÄÄ �ÄÄ0

Apparently mkvextract cannot convert subtitle format/codec, but just gives you whatever is in the Matroska file.

Going to try the ffmpeg suggestion. Usually ffmpeg is good about honoring the file extension you provide, if I remember right.

And just in case there are any doubts, here are the actual lines from the command-line session, except I've changed the filenames:

$ mkvmerge -i video.mkv
File 'video.mkv': container: Matroska
Track ID 0: video (MPEG-H/HEVC/h.265)
Track ID 1: audio (AC-3/E-AC-3)
Track ID 2: subtitles (HDMV PGS)
Chapters: 14 entries

$ mkvextract tracks video.mkv 2:subtitles.srt
Extracting track 2 with the CodecID 'S_HDMV/PGS' to the file 'subtitles.srt'. Container format: SUP
Progress: 100%

The lines of binary gibberish above are from subtitles.srt.

---

UPDATE: Nope, ffmpeg can't do it either. It notices the .srt extension and provides this helpful error message:
Subtitle encoding currently only possible from text to text or bitmap to bitmap

Hi I keep getting the message that Error: The file 'Episode_1.mkv' could not be opened for reading: open file error.

Me too, file is not open. or in use

Hi Guys,
This is not a support channel.
This is just drop-in "snippet".

If you need help, please move into more appropriate place where you can actually get some help.
Try https://stackoverflow.com/

Please remove me from this mailing list. I have no idea who any of you are. You have the wrong email address.

…

On Thu, May 7, 2020 at 5:48 PM Pavel Binar ***@***.***> wrote: ***@***.**** commented on this gist. ------------------------------ Hi Guys, This is not a support channel. This is just drop-in "snippet". If you need help, please move into more appropriate place where you can actually get some help. Try https://stackoverflow.com/ — You are receiving this because you commented. Reply to this email directly, view it on GitHub <https://gist.github.com/20a3366b54f41e355d2745c89091ec46#gistcomment-3293965>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFJKAEICMFEFIALXUDY3PT3RQJYULANCNFSM4IK3J2YQ> .

I did exactly as suggested. It extracted a file. The file ended in .srt because that's the filename I specified, but it's a 25MB binary file, not a plain text .srt file. I suspect it's a .sub file format. Here are the first couple of lines:

PG�9ÅÆ��������Ä�8���Ä��������˛�mPG�9ÅÆ������
���˛�m�á�îPG�9ÅÆ������B����ÄÄÄ��ÄÄ@��ÄÄ`��ÄÄø��ÄĘ��ÄÄ���ÄÄü��ÄÄ �ÄÄ0

Apparently mkvextract cannot convert subtitle format/codec, but just gives you whatever is in the Matroska file.

Going to try the ffmpeg suggestion. Usually ffmpeg is good about honoring the file extension you provide, if I remember right.

And just in case there are any doubts, here are the actual lines from the command-line session, except I've changed the filenames:

$ mkvmerge -i video.mkv
File 'video.mkv': container: Matroska
Track ID 0: video (MPEG-H/HEVC/h.265)
Track ID 1: audio (AC-3/E-AC-3)
Track ID 2: subtitles (HDMV PGS)
Chapters: 14 entries

$ mkvextract tracks video.mkv 2:subtitles.srt
Extracting track 2 with the CodecID 'S_HDMV/PGS' to the file 'subtitles.srt'. Container format: SUP
Progress: 100%

The lines of binary gibberish above are from subtitles.srt.

UPDATE: Nope, ffmpeg can't do it either. It notices the .srt extension and provides this helpful error message:
Subtitle encoding currently only possible from text to text or bitmap to bitmap

you solved this?

I use zsh on my Catalina and I'm getting:

zsh: command not found: mkvmerge

My solution: I had the GUI MKVToolNix installed (v47) (via DMG install), so I did like this:

/Applications/MKVToolNix-47.0.0.app/Contents/MacOS/mkvmerge -i myFile.mkv

and

/Applications/MKVToolNix-47.0.0.app/Contents/MacOS/mkvextract tracks myFile.mkv 3:myFile1.srt 4:myFile2.srt

Worked.

https://stackoverflow.com/ is better place for this type of support requests / discussion.

I have the same issue as @larryy and @antonioreyna. Any updates or solves?

I have the same issue as @larryy and @antonioreyna. Any updates or solves?

I was dealing with the same problem and what worked for me was just simplify the name of the mkv file, removing the brackets that it was and everything worked well

Works on M1 Mac mini. You just need to fire up Terminal in Rosetta 2 mode before remote installing brew. Don't forget to set a folder location for extracted SRT files. Otherwise it'll just save them under main User directory. Thanks for sharing.

If the tool is installed and in your $PATH it will run. The name of the file is irrelevant to the .srt subtitle problem. Unfortunately, there are multiple issues being discussed, not all having to do with subtitles.

That said, trying to extract text subtitles, like .srt, from a .mkv file that has bitmap subtitles won't work, whether from the command line or using a GUI utility like mkvToolnix. mkvToolnix will extract bitmap subtitles to a .mks file, and I suspect that's what this tool is doing as well. Either this tool or ffmpeg would have to implement OCR to convert bitmap subtitles to text subtitles. Sadly, neither one does, but it's kind of understandable, since open source OCR tools are not very good without a language model of some kind. There's a Mac application called Subtitle Extractor in the App Store that does this, but it has no language model, so it will make silly mistakes like replacing "silly" with "sil/y", "I'm" with "I 'm", "with" as "With", "won't" as "won 't" and on and on. It gets enough right that you could probably fix it by hand, but it would be tedious because there are a lot of errors. Better than nothing, I guess.

Consider it a complex feature request I guess, to implement OCR with a language model to convert bitmap subtitles to text subtitles.

Extract subtitles from MKV on all subdirectories

extract_subtitles.py

from os import walk
import subprocess
import re
from os import path

tool_path = "/Applications/MKVToolNix-51.0.0.app/Contents/MacOS/"
dir = "./"

def find_files(dir, ext):
    file_list = []
    for (dirpath, dirnames, filenames) in walk(dir):
        for filename in filenames:
            if filename.endswith(ext) and not filename.startswith('._'):
                file_list.append(dirpath + '/' + filename)
    file_list.sort()
    return file_list

for file in find_files(dir, ".mkv"):
    basename = file.replace(".mkv", "")
    
    if path.exists(basename+".srt") or path.exists(basename+".ssa"):
        print("Already Exist, skipping...", basename)
        continue

    # Find subtitle track
    result = subprocess.run([tool_path + "mkvmerge", "-i", file], stdout=subprocess.PIPE, check=True)
    
    # SubRip .srt
    srt_track = re.search(r'Track ID (\d+): subtitles \(SubRip/SRT\)', str(result.stdout)) 
    if srt_track:
        srt_track = "{}:{}.{}".format(srt_track.group(1), basename, "srt")

    # SubStation Alpha .ssa
    ssa_track = re.search(r'Track ID (\d+): subtitles \(SubStationAlpha\)', str(result.stdout)) 
    if ssa_track:
        ssa_track = "{}:{}.{}".format(ssa_track.group(1), basename, "ssa")

    if not srt_track and not ssa_track:
        print('No SRT track found!', file, str(result.stdout))
        continue


    # Extract SRT
    subprocess.run(list(filter(None, [tool_path+"mkvextract", "tracks", file, srt_track, ssa_track])), check=True)

print("Finished!");

then run on terminal like:

python3 extract_subtitles.py

same issue exactly with an srt file of over 10MB with binary data.
SAD!

There's a Mac application called Subtitle Extractor in the App Store that does this

Thank you for this @larryy ! I needed an SRT to translate subtitles into another language and needed precise time codes, this app does it!

For SRT it is ok but for SUP like Japanese and Chinese, seems OCR is needed.