HOWTO create SRT subtitles from VOB files

[bernardfrancois]

FIRST METHOD:

Use the ksubtitleripper package from synaptic.

SECOND METHOD:

If ksubtitleripper doesn't work for you, or you dont want to use it you can follow these steps:

Install the subtitleripper package from synaptic.
Change your working directory to the directory containing the VOB files. These files should not be encrypted.
Copy and paste the script below into a terminal window.



# NOTE: this script assumes there are no more than 9 subtitle tracks
# creates subtitle files for the 9 first languages
for lang in {0..9} ; do
(
# extract the given language and creates a subtitle file
tccat -i *.VOB | tcextract -x ps1 -t vob -a 0x2$lang > subs-$lang ;

# converts the subtitle file to an .srt file (if the file isn't empty)
if [[ -s subs-$lang ]] ; then

(
# convert the subtitle frames to seperate images
echo "LANGUAGE $lang: WILL NOW CREATE SUBTITLE FRAME IMAGES" ;
mkdir pgm-$lang ;
subtitle2pgm -o pgm-$lang/pgm-$lang- < subs-$lang ;

# use gocr to convert the subtitle frames to text
echo "LANGUAGE $lang: WILL START OCR ON SUBTITLE FRAME IMAGES" ;
pgm2txt pgm-$lang/pgm-$lang- ;

#convert the text to a .srt file
echo "LANGUAGE $lang: WILL NOW CONVERT THE TEXT TO AN SRT FILE" ;
srttool -s -w < pgm-$lang/pgm-$lang-.srtx > subs-$lang.srt ;
) ;

else
# removes the empty subtitle file (a languages that wasn't available)
rm subs-$lang ;
fi
) ;
done


Notes:

ksubtitleripper didn't work for me. I reported a bug regarding the issues I had with it:
https://bugs.launchpad.net/ubuntu/+source/ksubtitleripper/+bug/421470
Depending on the actual subtitles you want to copy, you may have to be the copyright holder of the subtitles.
Use this script at your own risk.
Feel free to ask any questions about the script, or to send me a personal message in case it would be broken.

Note that

[legatek]

LANGUAGE 0: WILL NOW CREATE SUBTITLE FRAME IMAGES
LANGUAGE 0: WILL START OCR ON SUBTITLE FRAME IMAGES
Using /usr/share/subtitleripper/gocrfilter_none.sed to filter gocr output
creating directory ./db/
creating empty file ./db//db.lst
File pgm-0/pgm-0-*.pgm not found
File pgm-0/pgm-0-*.pgm.gz not found
LANGUAGE 0: WILL NOW CONVERT THE TEXT TO AN SRT FILE
/home/monkey/subtitle-extract.sh: line 23: ../../subs-0.srt: Permission denied

[legatek]

I tried on another movie, and this was the console output:

LANGUAGE 0: WILL NOW CREATE SUBTITLE FRAME IMAGES
LANGUAGE 0: WILL START OCR ON SUBTITLE FRAME IMAGES
Using /usr/share/subtitleripper/gocrfilter_none.sed to filter gocr output
creating directory ./db/
creating empty file ./db//db.lst
File pgm-0/pgm-0-*.pgm not found
File pgm-0/pgm-0-*.pgm.gz not found
LANGUAGE 0: WILL NOW CONVERT THE TEXT TO AN SRT FILE
LANGUAGE 1: WILL NOW CREATE SUBTITLE FRAME IMAGES
LANGUAGE 1: WILL START OCR ON SUBTITLE FRAME IMAGES
Using /usr/share/subtitleripper/gocrfilter_none.sed to filter gocr output
File pgm-1/pgm-1-*.pgm not found
File pgm-1/pgm-1-*.pgm.gz not found
LANGUAGE 1: WILL NOW CONVERT THE TEXT TO AN SRT FILE

-----

db, pgm-0 and pgm-1 directories were created, and the pgm directories had files named pgm-0-.srtx and pgm-1-.srtx, respectively, both with size 0.

[bernardfrancois]

Was this a VOB file from a commercial movie? Typically, those are encrypted. Probably my script doesn't work on encrypted VOB files.

If anyone tried with a non-encrypted VOB file, please let me know if it worked.

[phenest]

I've been having trouble decrypting a copy protected DVD (Alice in Wonderland & Avatar). Then I found this script...

I've been ripping from DVD's (even copy protected) using...
mplayer dvdnav://$title -dumpstream -dumpfile $name.VOB
...and discovered this script works on that file. Nice.

One question: when it asks to enter a character it doesn't recognize, how does one enter italic or bold? Apparently I can enter it as UTF8, but I don't know how.

[phenest]

Ok, so you can't enter italic or bold characters because gocr doesn't support it.

[maghoxfr]

Isn't it possible using Avidemux?

[andrew.46]

Hi,

I see that the writer of this guide has not been active on these forums for a while which is a great pity. I have not used this script but I have certainly used the tools mentioned in it and I can add one small snippet to the pgm2txt syntax. If you have an image viewer defined in the script /usr/bin/pgm2txt alter the line to:

pgm2txt -v pgm-$lang/pgm-$lang- ;

and an image viewer should then show the exact image that gocr will query when the OCR fails to guess the text. Makes life a lot easier :).

Andrew

[phenest]

Ok, so you can't enter italic or bold characters because gocr doesn't support it.

It may be possible as gocr allows a string to be entered for each character, rather than just one character. This means you could theoretically enter "<i>P</i>" for an italic P. But it seems to double up these strings in the finished text. Not sure why, maybe a bug? I'm having a look at gocr's code to see why. If this theory worked, it would only require a script to remove surplus formatting tags as each character would be formatted individually. Incidentally, the gocr package in Debian Squeeze is nearly 3 years out of date, and I'm guessing Ubuntu's will be too. There is a much more up to date source code available from their site.

Isn't it possible using Avidemux?

Only for hard coded subs I believe.

Thanks for the tip Andrew.

[maghoxfr]

Only for hard coded subs I believe.

Thanks for the tip Andrew.

You got me, I wasn't aware of that.
See you around

[gecka]

pgm2txt is a total fail it cannot convert simple numbers like attached