[Vorbis] Vorbis at first place in "Public, Multiformat Listening Test @ 128 kbps"

Balatoni Denes dbalatoni at programozo.hu
Wed Jan 18 11:04:39 PST 2006


szerda 18 január 2006 01.36-kor Tor-Einar Jarnbjo wrote:
> Maik Merten wrote:
> >The setting were chosen so that all encoders will give 128 kbps on
> >average with *full length* tracks/albums (which is what is interesting).
> I doubt that. All encoders exceeded the target bitrate substantially,
> the Vorbis encoder at most with almost 16%. When testing specific audio
> material, they should also tweek the encoder quality settings, so that
> they reach the target bitrate with the applied settings.

For more information, look at the pretest discussion (the settings were indeed 
choosen so that a bitrate of ~130kbps is achieved on a larger amount of 

> >Thus the 7-8% bitrate difference during these short test-samples is
> >uninteresting as there are enough sequences in which Vorbis will go
> >below 128 kbps.
> And the other codecs won't go below 128 kbps on those sequences, so that
> the average difference remains? What I got to think about is wether or
> not the Vorbis header was included in the bitrate calculations? That
> might of course explain the difference on such short samples.

It wouldn't, the header amounts to 1-3 kbps on these short samples (cca. 3 
kbytes in the file). But indeed, Vorbis is probably willing to go lower 
bitrate wise than the other contenders.

> Tor


What kills me, doesn't make me stronger.

More information about the Vorbis mailing list