[mythtv] Commercial Detector questions

f-myth-users at media.mit.edu f-myth-users at media.mit.edu
Sat Dec 10 02:52:02 EST 2005


    Date: Fri, 9 Dec 2005 17:16:48 -0500 (EST)
    From: f-myth-users at media.mit.edu

    Another idea that might deserve some thought:  Looking for the sudden
    disappearance of closed-captioning information.

If anyone's inspired to actually try this, I just dug up a really
excellent tool that will help; it was mentioned a couple months ago
on the ivtv-devel list and I just found it.  Thanks to Hui Zhou for
writing this:

http://www.wam.umd.edu/~zhouhui/vbiutil-20050928.tar.gz

This tool produces timestamped ASCII text of the closed captioning,
suitable for input via the -sub switch for mplayer (assuming you've
made sure mplayer has OSD fonts and so forth).  Because it's indexed
to the stream, a commflagger should have no trouble noticing gaps in
the text and correlating it with whatever other visual processing it's
doing.  [Note that it assumes the format produced by ivtv (and maybe
bttv?).]

This tool probably also solves another problem of mine, namely that
Myth doesn't preserve CC data when transcoding to MPEG4; it looks like
I'll be able to just preserve the CC data from MPEG2 in a separate
file and have mplayer redisplay it directly in the X video stream it's
generating (e.g., not by trying to embed CC data in NTSC line 21, as
the PVR-350's native MPEG2 output would do).  This is a little klunkier
because it requires frobbing mplayer to turn the CC on & off (and I
don't know if this can be done in mid-stream, so to speak), unlike
just using the TV controls to do it, but it'll probably do in a pinch.
And for machine-searchability it's actually much better to have this
data in a separate file from the much-larger MPEG anyway.

[Now my only problem is that the transcoder seems to be producing
files that have no audio and blow up mplayer after a few seconds;
that's a separate bug report.]


More information about the mythtv-dev mailing list