switchable subtitles - best solution?

[tripex]

Since years I subrip my DVDs with SubRip to srt files, for each language one beside the main avi. This works good so far but I consider to change my habit towards more advantages, if there are any.

The highest priority is accuracy or in other words I want a minimum of correction work after the OCR process. DVDSubEdit is a fast tool (no OCR learning), but overall it takes longer to correct all the mistakes (words together) than using SubRip and learn every single letter and correct only a few mistakes after. Are there other (better than SubRip) OCR processing tools available? I don't conserve italic, bold or other fonts but accents like á or à is a must.

However basically I have an attitude to conserve as much as possible from the original. Therefore I never liked the OCR processing too much. It has some advantages like very small filesize and is highly editable (i.e. to remove hearing impaired), but has also heaps of problems. First the process to create takes just too long, second you can never be sure, there are no OCR errors until you check EVERY single sup. Third you have problems with special characters and asian languages are nearly out of option at all.

As you see I tend to a graphic format and obviously the most supported is idx/sub. I've got it with VSRip. My BSplayer and others can handle it straight out of the box very well (without VobSub) but I encountered one big problem. It looks awful compared to srt. With VLC it becomes even worse so I could not recognize it. How can I improve the rendering quality OR switch of the resizing at all? As I know the sups on DVD and idx/sub are saved as 720x480 bitmaps. It should be fine to watch them without any resizing at 1024x768 fullscreen.

Secondarily it consumes about 3 MB for each language, but I realized the sub is highly compressible. Is there a way or maybe another format with integrated (de)compressing mechanism?

Last, what is the best tool to post-edit idx/sub files to remove (not hiding) unwanted sups or even to alter sups? I know, that I can do it with DVDSubEdit, but only with VOBs

CU
Andy

[manusse]

You can edit VobSub (sub/idx) subtitle either with the original VobSub package: VobSubStrip, SubResynch or with SubtitleCreator. However the editing is limited to timing adjustments, color change, position change.

You can also use SubtitleCreator to convert your subtitles to the Sup format accepted by DVDSubedit. Then you can edit them with DVDSubedit.

Regarding the size of the subtitles, some players (MPlayer,) accept sub/idx files compressed in a single rar file of the same name (with .rar extension).

Cheers
Manusse

[jeanl]

If you find that words have a tendency to be glued together, you can adjust the "Word boundary detection" option in the preferences of DVDSubEdit to tweak it.. Do some experimenting. It might minimize the number of subtitles you have to correct by hand afterward...

http://img88.imageshack.us/img88/7229/03022007220822yi3.png (http://imageshack.us)
Jeanl

[setarip_old]

@jeanlyou can adjust the "Word boundary detection" option in the preferences of DVDSubEdit to tweak it.Is it that, if "2" is selected, the program presumes that a group of two or more consecutive blank spaces equals a word boundary - and if "4" is selected, the program presumes that a group of four or more consecutive blank spaces equals a word boundary?

If that's not the case, please explain what the number choices represent.

Thanks

[jeanl]

Well, there's no real good explanation!!! It's a parameter that controls how sensitive the word detection algorithm is. The smaller the number, the easier it is for the algorithm to separate characters (so the more words you'll get). The higher the number the fewer words you'll get... But it's not very scientific. It's a parameter inside GOCR, that I just expose to allow people to try various settings...
jeanl

[setarip_old]

The smaller the number, the easier it is for the algorithm to separate characters (so the more words you'll get). The higher the number the fewer words you'll get...Thanks for the insight ;>}