[mythtv] Broken AV sync after mythtranscode

[Adam Egger]

On 12/22/05, Geoffrey Hausheer <mythtv0368 at phracturedblue.com> wrote:
> On 12/22/05, Adam Egger wrote:
> > The same problem here, I'm using today's svn (system: debian sid,
> > dvb-t, Germany, kernel 2.6.14). It looks like it's starting to get
> > asynchronous after there were some bad frames in the stream. I'll send
> > you the two logs (the -v all log and the whole mplayer log for this
> > file) soon.
>
> > Used command:
> > mythtranscode -l -c 9 -s 2005-12-18T14:28:00 -o test.mpg
> > --showprogress -v all > log.txt
> > By the way, --showprogress doesn't work for mpeg2fix
> I'm looking at your logs, but see only one bug (caused by mythtranscode):
> in mplayer:
> A:1599.9 V:1599.9 A-V:  0.002 ct:  0.171 39992/39992 10%  3%  0.7% 15
> 0 0%
> A:1599.9 V:10593.6 A-V:-8993.666 ct:  0.167 39993/39993 10%  3%  0.7%
> 15 0 0%
> A:1600.0 V:1600.0 A-V:  0.002 ct:  0.167 39994/39994 10%  3%  0.7% 15 0
>
> In mythtranscode:
> 2005-12-22 13:28:25.423 VID: I #:0 nb: 2 pts: 00:26:39.706 dts: 00:26:39.666
> 2005-12-22 13:28:25.423 Id:0 00:26:39.706 V:181 MP2: 173
> 2005-12-22 13:28:25.423 VID: I #:2 nb: 2 pts: 16:11:55.271 dts: 00:26:39.706
> 2005-12-22 13:28:25.423 Id:0 16:11:55.271 V:180 MP2: 173
> 2005-12-22 13:28:25.424 AUD #1: pts: 00:26:39.108
> The 'Id' which has a PTS of 16:11:55 is bad, and would likely have
> caused an issue if it wasn't at the very end of the stream.  I'll look
> into fixing it.
>
> I see some other stuff, which isn't good though:
> 2005-12-22 13:27:39.398 Id:1 00:13:18.348 V:139 MP2: 98
> 2005-12-22 13:27:39.398 Found invalid audio PTS (off by 00:00:00.504)
> at 15:58:33.424
> 2005-12-22 13:27:39.398 AUD #1: pts: 00:13:17.940
> 2005-12-22 13:27:39.398 Id:1 00:13:17.940 V:139 MP2: 97
> A backwards moving PTS (from 13:18.348 to 13:17.940) will result in
> A/V desync.  This isn't a transcoder issue, and I don't really know
> any way to solve it.  If your stream is this broken, there isn't much
> the transcdoer can do.  I don't see why the transcoder would make it
> any worse than it already was though.  I don't see anything in the
> mplayer logs telling me about an A/V desync either except for this:
> A: 798.3 V: 798.3 A-V: -0.005 ct:  0.553 19952/19952 10%  2%  0.7% 6 0
> 43%
> A: 797.9 V: 798.3 A-V: -0.426 ct:  0.549 19953/19953 10%  2%  0.7% 6 0
> 45%
> which corresponds to the above backwards moving PTS.  It is about 0.5
> sec off, and mplayer corrects for it within 2 seconds.  So I guess I
> need a more explicit explanation of what you are seeing, and ideally,
> a clip around the area of interest that shows it in action.

The original file is ok. The transcoded file is quite big, so it's
rather hard to find the exact position where it starts to get
asynchronous. I'll try to find a smaller file with the same problem.

Thank you for your great work on mpeg2fix.

Adam