[Vorbis] Re: buggy vorbis encoder

Ben Collver collver at peak.org
Wed Dec 6 06:44:58 PST 2006


xiphmont at xiph.org wrote:
> On 12/5/06, Ralph Giles <giles at xiph.org> wrote:
> ...I'm not sure why it thinks a serialno of 0 is a problem, but that's
> not what's causing the trouble.

It is ogginfo that reported that a serialno of 0 could be a problem. 
Since they still played poorly after we changed the serial number, I 
don't think it matters to the Cowon decoder.

> I'm guessing the decoder in the Cowon is not tolerating the fact that
> the earliest Vorbis encoders could mistakenly put audio data on the
> same page as the final header packets.  This is not allowed in the
> spec, it was simply a bug in the oldest encoders.  Most playback tools
> don't care (in some ways, it requires extra work to not tolerate this
> bug)...
> 
> Any editing that causes a remux without a reencode (eg, the
> vorbiscomment example given earlier) will fix the problem by moving
> the first audio packets off the final header page.

On these songs where track 1 had a serial number of 0, I tried the 
vorbiscomment example, but they still play choppy on the Cowon G3. 
Since there are only a handful with this particular problem, I will 
reencode them.

Thank you,

Ben


More information about the Vorbis mailing list