Subtitle Software

[Roonan]

Can anyone recommend any software to extract and convert to text (SRT?) subtitle streams from Blu-ray and DVDs?

Thanks in advance.

[chip75]

Originally Posted by Roonan

Can anyone recommend any software to extract and convert to text (SRT?) subtitle streams from Blu-ray and DVDs?

Thanks in advance.

SubtitleEdit will do the job (you may have to extract them from your BDs or DVDs), but you may be better served looking for the .srt files online themselves. You're going to be going through checking the results either way.

You probably know that the subtitles on discs aren't text they're images, so the software guesses what each word is with OCR and converts them to text, sometimes its easier to find the end result than do the extracting/conversion.

[Naiera]

Yeah, much easier to just troll the interweblines for subs instead of OCR scanning. The only time I've ever done that is when a Hong Kong region 3 DVD of a German movie was the only way to get subs I could understand for the Blu-ray release

[Pyoko]

SubtitleEdit with the latest Tesseract engine (the program can download it for you before starting the OCR) will get you 95% there for regular subtitles, fully-automated. Of course that's also a double-edged sword as there's often a higher risk for errors compared to the more traditional semi-automatic OCR mode, where you enter characters manually.

Fixing and perfecting the last 5% is always the part that takes the most time, if you decide to do it. Stuff like OCR typos or confusion, merging or separating duplicate/overlapping lines, styles and positioning etc.

For more advanced subtitles SubtitleEdit has one major disadvantage which is that it assumes all subtitles to be on the lower part of the screen and will throw any positional information out the window, which is a problem for things like anime where they frequently put for example sign subtitles next to the actual sign in the image.

For this type of subtitle I've sometimes used SubExtractor 1.0.3.2d which is much older, simpler and cruder, and only has semi-automatic OCR, but it's often better than SubtitleEdit's semi-automatic mode (particularly with italics) and most importantly can keep positioning. On the other hand SubExtractor sometimes fails completely on certain types of weirdly colored subtitles.

[Roonan]

Thanks for the help so far. My main purpose was for audio commentaries. However, the ones I have from the Blu-ray are in italic and aligned to the left, which may complicate things during the OCR process. I'll give SubtitleEdit a try.

[Pyoko]

That sounds like the perfect case because presentation and complete accuracy is probably not hugely important. For me I've found these settings to be the optimal workflow:



It'll be 100% automated and usually successfully differentiates between regular/italic subtitles. Afterwards back in the main program window you can run an additional spell check pass to manually fix any uncertainties.