Extract only the timings of Japanese subtitles, to make a translated fansub

[yojimbo2000]

I have a Japanese DVD that I want to make English fansubs for (using Jubler).

The DVD has Japanese (for hard of hearing) subtitles. So to save myself the time of having to spot all the subs, I thought I'd use the timings from the Japanese language subs.

However, when I use D-Subtitler, it attempts to run the Japanese text through its OCR engine. Although it seems it might be possible to teach the OCR to recognise kana, I'm not typing in hundreds of kanji.

Assuming Japanese subtitle OCR is impossible (it would be wonderful if it was possible though... is it possible?) I just would like the timings. What I imagine is a dummy .srt file that I can load into Jubler, and replace the garbage/blank text with the lines of my fansub.

Any suggestions would be massively appreciated, I'm fine with command line stuff, Mac friendly if possible, but I could go to virtualisation if needs be.

Thank you for reading

[lowellriggsiam]

If you extract the subtitles as a sub file, you could open it in subtitle workshop.

[yojimbo2000]

Ok, I see what you mean, so it's extracting timecodes + bitmap images, and not trying to do OCR.

is sub the same as VobSub?

I did a VobSub extract using ffmpegX and I now have a .sub file (which is the images) and a .idx file (which is the timings).

I tried loading the .sub file into Jubler, and it seems to be working (Jubler has even had a go at OCRing the .sub images - it's just garbage). The important thing is that the timings seem to be there.

I'll need to convert the VIDEO_TS folder into an mp4 in order to test whether those timings are correct.

I'll let you know if it works.

[AlanHK]

Sorry, I'm a PC user, but presumably there are equivalent apps on Mac:

However, when I use D-Subtitler, it attempts to run the Japanese text through its OCR engine. Although it seems it might be possible to teach the OCR to recognise kana, I'm not typing in hundreds of kanji.

Assuming Japanese subtitle OCR is impossible (it would be wonderful if it was possible though... is it possible?) I just would like the timings. What I imagine is a dummy .srt file that I can load into Jubler, and replace the garbage/blank text with the lines of my fansub.

So let your program try to OCR it. Set it on full automatic. Doesn't matter what gibberish the text is, the timings should be correct.

Then open it in your subtitle editor and overtype your new text.
It might be nicer if you can delete al the gibberish (but leave one symbol on each line to remain valid).

You can do this if your subtitle editor has wildcard search and replace; if not open the file (as text) in a text editor and do it there if you can work out the format so as to preserve the times.

I've done things like that by saving the file as MicroDVD sub format, which looks like this:

{1}{1}25.000
{110}{193}Status of this film archive|does not meet the accepted
{198}{279}standards of quality,|but is included in the film
{281}{333}because of its uniqueness.
{430}{501}Cinematographic office|of Ministry of Culture

This is convenient as each timing and subtitle is one one text line.

So it's then easy to delete all on the right of the last } after the time codes. (Leaving the first line, which is the fps.)
E.g., using "column mode" in your wordprocessor -- take care as the number of digits changes when you come to 1000, 10000, 100000 frames.

{1}{1}25.000
{110}{193}A
{198}{279}A
{281}{333}A
{430}{501}A

The I can open this in the Subtitle editor and type the new text in.

An easier way, using Subtitle Workshop 2.51 on Windows:

Open your OCRed file in it, then do:
Edit/Translation/Translator mode.

This will create a new file with the same timings, each line with the text "- Untranslated subtitle -".
You can save the new file, and use as is, or S&R the default text to something shorter.

[yojimbo2000]

My mistake, it turns out Jubler doesn't load vobsub (.sub/.idx) files properly.

So. following AlanHK's tip, I loaded the .idx file into a text editor, and using column mode (plus an insert line number macro) eventually got the .idx to resemble an .srt. Just checked it in Jubler and it worked perfectly.

Thanks for all the advice!