[Vorbis] metadata

Ian Malone ibm21 at cam.ac.uk
Thu Nov 17 04:06:27 PST 2005


Ian Malone wrote:
> Ian Malone wrote:
> 

>>
> 
> Helix can be fixed by creating a dummy codec for the demuxer
> to throw packets at:
> <http://lists.helixcommunity.org/pipermail/xiph-dev/2005-November/000166.html> 
> 

As suggested on the Helix/Xiph mailing list seeking is broken
by that fix when the unknown stream is mixed with data pages
(see below for the test file used).  I'll be looking at solving
that soon hopefully.  The good news is that GStreamer (Totem
and Kaffeine) seeks okay (I managed to trip Totem once, but I
don't know if that's related).  Rhythmbox sometimes has trouble
adding XMLed files to its library, whether this is a Rhythmbox
bug or it's using GStreamer functions that Totem and Kaffeine
don't I haven't investigated.

Mplayer seeks with no trouble.  Xine seeks, although it has a
slightly odd pause right after a seek, I should compare that
with normal Ogg/Vorbis seeks.  XMMS (vorbisfile based I think)
seeks fine, but glitches (short, with varying severity) during
play.

> 
> I've also gotten hold of one of the pan.zipcon.net files and
> XMLed it up:
> <http://www.srcf.ucam.org/~ibm21/Bach-Busoni_Nun_freut_euch_3.rdf.ogg>
> and the XML as a separate file:
> <http://www.srcf.ucam.org/~ibm21/Bach-Busoni_Nun_freut_euch_3.xml>
> 

And now the same file with the description packet repeated
every five Vorbis pages (this is what following the XMPP
approach to streaming looks like as far as I can tell):
<http://www.srcf.ucam.org/~ibm21/Bach-Busoni_Nun_freut_euch_3.multi.ogg>

I can post the code used to create these streams if anyone
wants, but it's fairly trivial and my approach is a bit
hackish.  It's also turned out useful to have a stream
checker tool which gives page numbers, granulpos and
serial numbers; is there anything like this under svn
already?

I'm linking test files at:
<http://www.srcf.ucam.org/~ibm21/ogg_meta_data.html>

There should probably be a bit more exposition on the
status of different players, but I was mainly concerned
with not putting people off Vorbis when they found a file
they couldn't play.  (Since only about two people other
than me have downloaded the first test file I probably
shouldn't have worried)

Apologies to anyone getting bored of this, I'm mainly using
it to keep track of what's been done.  Maybe it belongs more
on Vorbis-dev?  There's not much 'dev' being done with this
though...

-- 
imalone


More information about the Vorbis mailing list