[vorbis] quality question?

Gerhard Thomas Kriz gkriz at edv1.boku.ac.at
Thu Jul 25 00:42:40 PDT 2002



Actually I have been doing a very subjective test yesterday, 
encoding a wav (Dub Syndicate: "Creation") I am very familiar with 
and I think of being a bit demanding (Dub Syndicate: "Creation") to 
oggs, using q4 up to q8 and q10. After comparing the sound (= 
listening very carefully)I was decoding those files to wav again and 
analyzed them (I was using WinVorbis 1.35 for encoding and 
OggDropPXd for decoding).
Personally I found that q4 up to q7 was altering the sound too much; 
q8 came very close to the original. I didn't listen to q10, because 
those kbps are far too much for my purpose.
Interesting enough, after decoding those files and analyzing them, 
there were clippings to be found in all except in the wav decoded 
from the q10-file. The q4-wav had the most clipping, the q8-wav the 
least. 
I guess I am going to stick with q8, although I was hoping to get 
away with a much smaller file, means q5 or q6.
Maybe this is helpful to other newbies...
Tom.

> On Thu, 25 Jul 2002 18:01, Robert Cole wrote:
> > No one has any idea on my question of recompressing ogg files?
> 
> If you can tell the difference between -q 7  and anything less, then you may 
> introduce artifacts you can hear by re-encoding down to -q 7 from -q 9. If 
> you can't, you might get away with it - try it, and trust your ears. Better 
> to either re-encode form the original source, or wait out for the coming of 
> the legendary bit peeling tool, though. 
> 
> Personally, I like to try ABX tests on a collection of samples, such as the 
> various tricky ones used for codec testing, and music I'm very familar with. 
> I try out various quality levels looking for a sweet spot, where I can no 
> longer distinguish the encoded form from the original, then add about 0.5-1 
> to account for future improvements in sound gear, ear training, and those 
> tricky to encode peices that I don't know about yet. 
> 
> Of course, if you're not fussy about storage size, pick something high, but 
> bear in mind that the silly bitrates will almost certainly receive some extra 
> tuning attention in the next 6 to 12 months.
> 
> 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.

<p>Thomas Kriz
Public Relations / BOKU Wien
University of Agricultural Sciences
Gregor-Mendel-Strasse 33
A-1180 Vienna / Austria / Europe
Tel. ++43 1 / 47654 1017
Fax  ++43 1 / 47654 1016

--- >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.



More information about the Vorbis mailing list