[vorbis] Pitch shift with RC2

[Greg Wooledge]

Moritz Grimm (gtgbr at gmx.net) wrote:

> Greg Wooledge wrote:

> > I heard it on one sample, which I've already sent to Monty.  The
> > hi-hat cymbals in one song sound significantly lower in pitch when
> > I encode with Ogg at "-b 128".  I can't hear it at "-b 160", though,

I've put this sample up at http://wooledge.org/~greg/slave.wav.flac (438k).
It's only 3 seconds; if someone wants a longer sample, let me know.

> I ogged my problematic bassrumble tune at all available bitrates, and
> since I know my tune best I think I would have noticed such a thing.

Hmm, we're probably talking about two different things, then.  A bass
rumble and a hi-hat pitch drop may not have the same root cause.

> On the other hand, the compiler can be a
> problem source - OpenBSD 2.8 would crash on high FTP network traffic if
> the kernel was compiled for CPUs > i386.

I compiled the 1.0rc2-1 debs on Debian unstable, using gcc 2.95.4-4.

> Maybe this is some rather random problem that depends on what configure
> chooses as parameters that are used within the Makefile, depending on
> CPU (propably AMD vs. iNTeL? I am free from trouble with my Athlon here)
> or so.

I compiled and encoded on a K6-2, but since the .deb was built with
all the default options, I doubt that matters.  In fact, I installed
the same .deb on a Celeron box and used it there, so I really doubt any
K6-specific optimizations were used. :-)


-- 
Greg Wooledge                  |   "Truth belongs to everybody."
greg at wooledge.org              |    - The Red Hot Chili Peppers
http://wooledge.org/~greg/     |



<HR NOSHADE>
<UL>
<LI>application/pgp-signature attachment: stored
</UL>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: part
Type: application/octet-stream
Size: 241 bytes
Desc: not available
Url : http://lists.xiph.org/pipermail/vorbis/attachments/20010816/3edc4c0a/part-0001.obj