[vorbis] That pesky udial.wav again...

[John Morton]

I finally got RC3 built on my home machine (had to sneakernet in a tarball of 
libcurl to get it to build :-( ) and started testing it out on the range of 
quality levels, side by side with lame using it's range of vbr quailty modes. 

Overall, it sounds pretty good, but I haven't really tested a good range yet, 
as I ended up starting on udial.wav... vorbis handles it ok up to quality 3, 
but from 4 onwards (presumably a high frequency cutoff is removed) there are 
very obvious artifacts, and they continue to be present all the way out to 
10.0. Lame handles this a lot more gracefully - the sine sweep is cut of 
until about V2 or 3 (but the hiss is still present), and it sounded pretty 
close to the original at V0 (though I could still got 12/16 in an ABX test, 
the difference was barely perceptable). 

Now I _know_ that high frequency sine wave sweeps are a completely contrived 
sample to test an audio codec with, but I'd hate to think that some idiot kid 
on slashdot will end up badmouthing vorbis because it goes to pieces on a 
sample like this. It would be a lot nicer if it failed to include the sweep 
and didn't contain artifacts. 

That said, fatboy.wav and hihat.wav where sounding pretty sweet most quality 
levels. Nice work :-)

John

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.