[vorbis] what quality ?

Moritz Grimm gtgbr at gmx.net
Sun Apr 21 04:26:36 PDT 2002



Rok Pape¾ wrote:
> > Phase Stereo:  Makes use of the fact that the human ear is nearly deaf
> > to signal phase above 4kHz, meaning that the ear can't tell very well
> Above 4kHz? Shouldn't this be below 4kHz?

No. Below 4kHz, humans start to hear pretty good, because that's where
the frequency range of the human voice starts (roughly). Signal phase is
important for us in that frequency range, since in addition to the
different volumes on the left or the right ear, this is how we
"calculate" the position of a sound source pretty accurate - we measure
the time between when a wave reaches the one and then the other ear.
Above 4kHz, waves are getting so small that the phase shifts get bigger
than 360° - we don't have much use for that, which is why we mostly
judge the position of the sound by the different volumes we get. Our
head takes a lot of the energy of a high sound while traveling to the
second ear, which is why that works pretty well. Our body is like a
lowpass filter, and we make use of that (AFAIK ... i'm not entirely sure
about that).
We usually fail to locate bass, because a wave of these frequencies is
much longer that the distance between our ears. The phase shift is too
little to be really noticed, and it gets worse the lower the sound
becomes. In the real world, bass is more or less all around us, it
penetrates us - those waves hardly get damped by any of our body parts
(our body's lowpass cutoff is above those frequencies), which is why we
don't percieve any significant changes in volume.
This, btw, is the reason why panned and/or heavily phase shifted bass
sounds so weird on headphones. It's unnatural. In addition to that,
phase shifts around 180° in the bass area interfere a lot - they
noticeably delete each other, that's why wrongly connected speakers
sound so bad.

So, as far as I understood all that, leaving out stereo information in
those areas where we can't percieve them anyways (or limited, as with
the high frequencies), is pretty useful in saving space. ;)

I find this topic pretty interesting, so if I said anything wrong,
please correct me.

<p>Moritz

--- >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