[vorbis-dev] some listening tests

[Aleksandar Dovnikovic]

<ben at slumber.dhs.org> wrote:

> I have a couple of samples that produce interesting artifacts when
> encoded with the CVS snapshot of 2001-02-13.  Both are about a meg.
> ftp://slumber.dhs.org/tmp/4.wav.bz2
> When encoded with oggenc -b 128, there is a sort of stereo separation
> in the drums, while in the original they are 'solidly' positioned in the
> stereo field.  This also occurs with -b 160 and is barely audible with
> -b 192.

I'm not one of the developers so my opinion could be wrong...
With this first sample, I think that the problem is in the pre-echo,
because although the drums are positioned in the centre, some
of the drum sound (less audible) also spreads more toward the
edges but pre-echo adds more noise to them so they become
more audible and this may look like wrong stereo separation
(I've noticed this also with one of my drum clips).

> ftp://slumber.dhs.org/tmp/beetles.wav.bz2
> 
> In this one there is a breathy vocal that becomes noisy and has the same
> stereo effect as above.  This is audible with -b 128 through -b 256.
> 
> In each of these, the part of the sound in which the artifact is audible
> is a panned mono sound mixed with stereo sounds, and I can't hear it
> with samples that are just a panned mono sound.

It could be that the background noise is not very well preserved so
it becomes more noisy; but since I haven't noticed this before it is
more likely that the psycho-acoustic doesn't work well on vocals
all the time - this is a known issue so some codecs take special
care when encoding vocals.

Greetings,
Aleksandar

--- >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-dev-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.