mpg2srt help - output ends after first min

[rutiger]

After much struggling with registry keys I've finally been able to get my PVR-350 to preserve closed captions when encoding. However, when I run mpg2srt I'm only getting the first minute of captions written to an .srt file. Any ideas as to what's wrong?

I'm on ver 0.2 - I understand that ver 0.3 is out, perhaps that might solve the problem?

[merrypig]

It's a hastily cobbled together webpage, but you can get v0.3 here ->

http://www.htpctools.com/mpg2srt/


My hunch is that your programs use a different captioning style that isn't supported in v0.2. v0.3 should fix that.

Regards

[rutiger]

wow - thanks for the quick reply.


0.3 gives me the entire transcript, although now there appears to be one line for every few thousandths of a second - 2 or 3 lines for every letter!

eg

4950
00:08:42,667 --> 00:08:42,700
OF A CONSERVATIVE GOVERNMENT.
AND TH
4951
00:08:42,700 --> 00:08:42,734
OF A CONSERVATIVE GOVERNMENT.
AND THAT
4952
00:08:42,734 --> 00:08:42,767
OF A CONSERVATIVE GOVERNMENT.
AND THAT I
4953
00:08:42,767 --> 00:08:42,834
OF A CONSERVATIVE GOVERNMENT.
AND THAT IS
4954
00:08:42,834 --> 00:08:42,900
OF A CONSERVATIVE GOVERNMENT.
AND THAT IS
4955
00:08:42,900 --> 00:08:42,967
OF A CONSERVATIVE GOVERNMENT.
AND THAT IS R
4956
00:08:42,967 --> 00:08:43,033
OF A CONSERVATIVE GOVERNMENT.
AND THAT IS REF

I may have to play around with recording settings some more. I'm currently using GB-PVR to record because its free and I'm still testing out the PVR-350, but if I get better caption results with BeyondTV I'll have to switch (although I'm still searching for a nice open source PVR package for XP).

R

[merrypig]

Yep. This is normal

The reason why you only saw some captions in the v0.2 file is that your tv show is using paint on captioning. That is, it adds 2 letters every frame. It's typical for news shows or live broadcasts where captions aren't pre-prepared. v0.2 did not support this format.

v0.3 does.

There are two ways to handle it; One, issue a new line in the caption file on each caption update. Or two, wait til the end of the sentence and then display.

The problem is, in the latter case the caption shows _after_ the audio is completed. So it's often too late to be useful imho.

I preferred the first option so that's what I coded in.

It should be transparent to you however when you view them. You'll see the captions building up like a teletype. Don't worry about the file size.

[rutiger]

I'm actually not using the .srt to display captions with the video - I'm trying to build a system that will archive the cc transcript and use it to create a searchable index of video content. However the .srt output is still very useful, only that it requires some work parsing out the redundant text.

I recall on your old mpg2srt site you mentioned that source code would be forthcoming - is this still in the works? I'd like to get my programmer colleage to work on outputting cc text via the second route you describe above.

r

[merrypig]

Ironically, I actually had started to clean up the code, release a new version and source ...

... and a power failure took out my pc and I've spent the entire week rebuilding my 3 systems here. I'm back online now, but missed my opportunity to knock the code out. Maybe this coming week.

I had a flag in my code to choose between the two output modes for paint on captions, I just never exposed it via cmd line. It should be trivial to customize it to suit you.

Better yet, I plan to extrapolate the caption and buffering code into a library with api so that you can easily link it into your own projects.

Just need the time to do it.

[rutiger]

Ouch - hope nothing important was permanently lost.

I'll follow this thread over the next week to see if you get a chance to dump out the code. Thanks again for all your help.

[merrypig]

Nah, nothing lost other than hardware. I just had to juggle my motherboards and memory around to fit the replacements - so it meant reinstalling a bunch of crap (different chipsets etc).

Stay tuned

[rutiger]

Hey merrypig - just wanted to see how things are progressing with releasing mpg2srt source code and/or exposing the flag to output captions a sentence at a time. Looks like it will take our programmer a fair amount of work to parse out the output we're working with now, so exposing this option will save us a whole lot of work.

[merrypig]

Had so much to do lately, haven't had time to address this yet. Sorry - soon hopefully.

[rutiger]

No worries - I'll stay tuned