[ExtractStream] Re: Splitstream Survey

[Dale Reed]

Roger Merchberger wrote:

> 
> > No, you're correct -- that's exactly what it does. It takes the
> > audio and video
> > streams and merges them together. If ExtractStream works correctly
> > and
> > reads chunks correctly, it'll output one audio for every video so
> > there's no
> > synchronization problems.
> 
> That's what I thought. So am I also correct in assuming that the raw
> tystream is *not* a standard MPEG2 -- or else why do we have to do
> all this crazy "extract - convert - knit" stuff?


I'm not an expert, but heres my opinion:

In a normal MPEG stream, there is not a 1-to-1 for Audio
video, because video takes up a lot more of the stream than
audio. Just look at your m2v and m2a files and see what
the ratio is. :) Also, because MPEG uses compression for
both the video and audio, I don't think there is a set
video-to-audio rate.

When the tivo records a movie, its getting the mpeg encoded
audio and video from two seperate sources (chips) at the rate
you have defined them to be recorded. It stores these intertwined
in the tystream file. To show the program, it simply reads the
tystream file back out, sending the audio to the audio player
and the video to the video player. Because its sending the streams
simultaneously to each chip from the same source, they stay in
sync.

The initial work done with tympeg has never had a sync issue.
It just doesn't fully process the full (raw) tystream file
all the time. Tympeg is the first (AFAIK) program that is
trying to solve the sync issue by acting like the tivo does
(without splitting the audio/video to seperate files and then
re-constructing them).


Dale