Please specify the versions of the relevant software on your computer, as on my computer there is no such message (i.e. the bug is a regression and I am still on the "good" end of it).

gstreamer=0.10.35
gstreamer-plugins-base=0.10.35
gstreamer-plugins-bad=0.10.22
gstreamer-plugins-good=0.10.30
gstreamer-plugins-ugly=0.10.18
totem=3.4.3

OTOH, the bug is not unique to totem, as mplayer also doesn't want to decode the audio without -demuxer lavf.

The file was prepared as follows:

1) Ripped the DVD: dvdbackup -M -i /dev/sr0
2) Copied the first video track and the first audio track using ffmpeg: ffmpeg -i VTS_01_1.VOB -ss 68 -t 10 -acodec copy -vcodec copy sample.mpg

It may well be that ffmpeg-0.10.4 muxes bad mpeg files such as this.

I'm using GStreamer 1.0.x (git master), it is a regression compared to 0.10, that's why I filed it :)

OTOH, as a developer of a DTS encoder (which, sadly, still has no gstreamer plugin), I would like to advise that you drop the libdca-based gstreamer plugin completely and switch to ffmpeg for DTS decoding if possible, thus maybe fixing (or, rather, hiding) this bug. Reason: all versions of libdca incorrectly decode all DTS streams that use the "perfect-reconstruction 32-band interpolation FIR" feature (not common on DVDs, but seen on some BluRays) due to a table that has been incorrectly copied from the specification. The corresponding fix in ffmpeg is http://git.videolan.org/?p=ffmpeg.git;a=commit;h=3fc45bdcf0031be526b5c1d66a9bbd386f816841 . Besides, there were some fixes for crashers in dcaenc SVN that never made it into any released version, and support for the XCH and XXCH extensions is only in ffmpeg, but not in libdca.

We should invert the priorities of avdec_dca and dtsdec.

> We should invert the priorities of avdec_dca and dtsdec.

I think we may first need to fix avdec_* to handle downmixing properly, including all the negotiation machinery. See bug #608892.

Can't reproduce this any more with that file and 1.12.4 or git master.