PDA

View Full Version : How to extract subtitles from an avi?


Topo Gigggio
21st January 2002, 22:41
Is it possible extract subtitles from an avi-divx which has them over videostream? I mean an avi which has them not in a separate file but impressed on it as part of the image.

Taric25
21st January 2002, 23:16
The only way of this possible is by OCR. I'm not sure of the exact settings though...

Linux
22nd January 2002, 00:39
If the AVI file is made with permanent subtitles you cannot remove them anyway. Say that you rip the subtitles with OCR scanning.
Then what? Do you want both permanent and removable subtitles in the same movie? :rolleyes:

Topo Gigggio
22nd January 2002, 15:48
The movie I'm talking about was made by someone I don't know and I do not like its quality, so I want subtitles in order to apply them on a different rip of the same movie. The movie is japanese and it's very hard to find italian subtitles of it.
What does it mean "by ocr"? Which program am I supposed to use?
Subrip exracts subtitles via ocr but only from vobs.

Linux
23rd January 2002, 01:53
OCR stands for Optical Character Recognition and is a usual technique in scanner software.
It does some trick by pattern match the characters against a database.
Either way Subrip and all other program need some assistance anyway.
I believe that you already have tested the programs in http://www.doom9.org/software2.htm#subs.
It is a lot easier to rip subtitle from a DVD since it is recorded in separate mpeg substreams.
Rips from AVI have to be done with a lot more help from the operator.
1. Where is the subtitle, exactly? Many movies have a few subtitles at odd paces (at the top etc).
2. Have the text absolute black background, or else there is a lot more problems for the ripper to see what character it is.
I think that if SubRip was able to rip subs from AVI you had to help it so much so its meaning got lost.
One thing you can do is to do it manually.
I suggest that you look through the movie with subs and write down when and what it says. Then make new subs from the text.

Topo Gigggio
23rd January 2002, 15:30
Thanks for your reply.
I did know the meaning of ocr, what I couldn't understand was how it works in video files. Is it sure that there is no application able to rip subs from an avi? The movie I have is subtitled in a black space just under film, so it makes theorically possible the operation.

Yusaku
23rd January 2002, 19:31
there is not and I doubt there will ever be something to do it. It would require quite some AI in addition to OCR and mainly there would not be enough users for this kind of software anyway.

I did this by going sub by sub in VirtualDub and writing frame numbers to SUB file (SUBs are better for this; but be sure to convert it to time-based format afterwards) together with text. Takes about three hours per hour of movie, so it is not that bad.

Another option for auto-timing is to use SSA's time from WAV feature, which works quite nicely... but it won't have SAME times :)

Topo Gigggio
23rd January 2002, 20:02
What is this WAV feature are you talking about? It could be a good compromise to me. Thanks :)

Yusaku
24th January 2002, 11:58
Get SubStation Alpha and read manual - it will generate times automatically from 8bit WAV file you provide (use VirtualDub to uncompress/strip WAV to 8bit)

Topo Gigggio
24th January 2002, 15:06
It sounds great! I will try it. Thank you! :-)