[vorbis] Re: encoding bitrates, different machines...

[Ed Sweetman]

On Thu, 2002-04-11 at 09:14, Jernej Simonèiè wrote:
> Hello Hugo,
> 
> 11. april 2002, 9:13:53, you wrote:
> 
> > Note the big difference in bit rate... both is RC3. My best guess at the
> > moment is that the former was with a -b setting while the latter was 
> > with a -q setting. I usually do everything at -q, but it might be that I 
> > forgot to do so with the former.
> 
> Of course there is difference if you compressed one set of files with
> -q and the others with -b. I remember seeing a bitrate graph shortly
> after rc3 was released of the same track encoded with -b and -q. While
> the -b graph was more or less around the average bitrate, the av.
> bitrate in the last third of track encoded with -q was almost twice the
> av. bitrate of the first part of track.

What you see with -b is a more conservative VBR.  Everything is closer
to the abr chosen.  This can be seen from
http://safemode.homeip.net/vbr-abr.png .  What it shows is that both
come out to the same average bitrate, but by choosing -b you get a much
more conservative vbr output than by choosing -q.  If -q is not buggy
and performs really good, then no, -b will not produce a file that
sounds any better than -q, almost all the time it will produce a file
that sounds worse at the same abr assuming you can tell the difference
anyway.  

o -b is abr, rc4 will introduce cbr and -q is vbr. All filesizes being
equal and the vbr routine not having any bugs, -q will always produce
the best file. 

I assume that -b automagically chooses stereo settings depending on the
difficulty compared to it's abr in which case it would be the same as -q
if the resulting abr is the same.  

<p>--- >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.