Extract CC with text positions

[Hakunamatata67]

Hi,

I have few MP4 files with embed ClosedCaptions.
I usually use CCextractor to extract the subtitles to SRT files, but I'd like to extract the CC with text positions. I tried to extract as SSA, but it won't work.

Is there any other program that can do it?

[Simon_Haddad]

I have been working on a program (a web page for now) that'll do exactly what you're thinking of. It requires a Closed Caption Disassembly file as input, and spits out a positioned SSA, formatted to look like closed captions.

I''ll put up some sort of tutorial on how to generate a disassembly file (CCExtractor to make a RAW file > raw2scc to make a scc file > ccasdi to make a disassembly file), but until then here's a script (bundled with a bunch of programs I'm really not supposed to be bundling and distributing) that'll do it automatically.

This assumes the source-file contains legit closed captions (cea-608) and not some other type of subtitles.


pre-pre-alpha

Edit: a lot of functions aren't implemented yet, like roll-up captions (used for live or near-live events) and paint-on captions (used when there's no time to build up a buffer, like immediately after a commercial break; NBC uses these on all their shows)

[Hakunamatata67]

Thanks, great program so far.
I successfully created an ssa file, but all subtitles duration are less than 1 second...

[Simon_Haddad]

Thanks, great program so far.
I successfully created an ssa file, but all subtitles duration are less than 1 second...

If you paste in some (or all) of your disassembly file, I can tell you what's wrong.

[Simon_Haddad]

It looks like the .raw file generated by CCExtractor is full of null bytes every other frame. (they show up as {} in the disassembly file; sent to me in a PM, for those others watching this thread) They don't add any data, but they do advance the timing of closed captions by one frame (≈ 1/30 second) each. Most recordings don't have these. CCExtractor does have code to fix this (you'll notice any .srt files generated by CCExtractor from a .ts file are properly synchronized), but then it technically wouldn't be a "raw" file if they did.


If your text editor has a find-and-replace function, replace all of the empty braces {} in your disassembly file with an empty string, and it looks like these subtitles will double in duration (starting earlier, but ending at the same time).

Sorry about the delay, but for some reason I didn't get a notification about your PM until just now.