Automate the conversion of subtitles from sup or sub+idx to srt

[mr.duck]

Has anyone managed to automate the conversion of subtitles from sup or sub+idx to srt like using a batch file for example?

Accuracy is not the most important thing. If it doesn't know the letters then do a best guess. But the possibility to train it for the OCR to be more accurate would be ideal. Once set up, it must run and complete the task with no help or prompting whatsoever.

Can it be done?

[netmask56]

Can it be done?

Maybe - but so far I haven't experienced any software packages that do - the closest in my experience for what I do is Subtitle Edit http://www.nikse.dk/se/

with this package I haven't had to do a lot of correction

[mr.duck]

That program is the most accurate OCR reader I've used (OCR via Tesseract)... when it works anyway. But I have many subtitles it can't read for some reason. It's best guess is a ~ for a whole block of text. Nothing strange or weird about that subtitle. The program works OK for most of them including dual line subs. Strange. Plus I can't find any info on how to use from command line so probably it's not designed for that.

Sort of off topic now, but I've found a way to use the subs by converting from sup to sub and avoiding OCR all together (I thought I would have to convert to text like to .srt). It's ugly but at least it works.

[jman98]

Sort of off topic now, but I've found a way to use the subs by converting from sup to sub and avoiding OCR all together (I thought I would have to convert to text like to .srt). It's ugly but at least it works.

Which "sub" format exactly is that? I've lost count of how many completely incompatible subtitle formats use "sub" as their designation. Some are bit maps. Some are characters. They're all different.

[mr.duck]

Which "sub" format exactly is that? I've lost count of how many completely incompatible subtitle formats use "sub" as their designation. Some are bit maps. Some are characters. They're all different.

The source was subs extracted from digital TV that end in .sup. It ends up as sub+idx I dunno much more than that. It is picture based for sure tho.