Regarding the limits of human hearing:

While the upper limit seldom is higher than 20 Khz I believe tests have shown that presence of overtones over 20 Khz can affect the way sound is percieved.

[way offtopic]
The music business is now pushing or trying to push the new formats sacd and dvd-a as the next best thing.
Not that I believe this needed because a well produced cd can sound very good and natural but that's another story.
Now also lot's of speakers manufacturers are developing so called hypertweeters or supertweeters which can produce very high tones, they say these are needed for sacd or dvd-a. Funny thing is that a speaker with such a tweeter can actually sound worse than a normal speaker because of the interference the super/hyper tweeter will cause with the normal one.
One tweeter that would be able to go from let's say 4khz to 50khz would be way better than the above system.
[/way offtopic]

This post has been edited by Bylie: Nov 1 2002, 18:43

funny. because cd playcbak requires a 20khz lowpass filter.

i read it somwhere. it was to cut out alaising artifacts or something like that.

nut now is i remind, i think i read it at r3mix.net

This post has been edited by sven_Bent: Nov 1 2002, 20:12

Yes, it was in r3mix.net.
I saw it too

but is it true that cd playback is done with a 20khz lowpass filter ?

from r3mix:
using a pure 22050Hz (44.1khz / 2) "full range" CD source signal is an illusion anyhow because the audio CD standard states you should use a 20kHz lowpass filter anyhow to avoid aliassing. Which is done in normal and professional audio equipment.

No, the lowpass is 22 kHz. I've measured tones that I burned and played in the analog output of the Yamaha CDX-860 CD Player, and recorded in 24 bits 96 kHz into the analog input of the Marian Marc 2. They were perfectly played, with maybe 1 or 2 db of loss, up to 22 kHz http://pageperso.aol.fr/lyonpio2001/cdr/cdr.htm

22 kHz lowpass is needed at the recording to avoid aliasing. It is performed at the playback, mostly digitally, to avoid feeding a waveform having digital "steps" into the ampli.

Frequencies above 20 kHz don't affect the way sound is perceived. No test has shown this. Oohashi's paper is often refered to in this matter as proving that overtones do affect the hearing. This is wrong. I don't have access to the paper, but it is quoted here : http://www.cco.caltech.edu/~boyk/spectra/spectra.htm

Results of the test :
1-people tested say they hear no difference with and without overtones
2-Their encephalogram is changed when the super tweeter is turned on
3-It doesn't return to normal when the tweeter is turned off.

Conclusion of the paper : overtones can affect the hearing even if people are not aware of it, and even when the tweeter is turned off : they have a durable effect.

I disagree, the right conclusion is just 1 : no audible effect, people said it themselves. Consequently, the effect showed in 2 doesn't affect the hearing. And 3 shows that 2 was not related to the super tweeter being turned on anyway.

This post has been edited by Pio2001: Nov 2 2002, 15:47

i thought that the 22.05 kHz lowpass was due to sampling? for discrete signals, frequency is periodic over omega = 2pi and even about omega = pi (where omega = 2pi/N, N = number of samples. you can set your sampling period to 1 s to make conversions easier), so that frequency is only single valued over the interval [0,pi].

edit: oh. nm. you were talking about 96 kHz signals...

This post has been edited by shimage: Nov 2 2002, 20:40

Your explanation is right if it is applied to the audio information that is stored into the digital file : only frequencies under 22050 Hz can be stored entierly and in a unique way.
Now, according to how the digital to analog conversion is performed, various results can be achieved. A bare digital to analog converter will hold the value of any saple until the next one , creating an output that looks like stairs. Stairs are in fact very high frequency noises, that are filtered with a lowpass : first the signal is oversampled in order to reduce the size of the steps, making the noise quieter, and higher in frequency, then the result is converted to analog, then a very high frequency lowpass filter is applied to remove the little noise.

Some people here knwo better than me and will correct me if I said something wrong.

In conclusion, if I understand well how it works, the oversampling process performs a natural 22050 Hz lowpass digital filter, isn't it ?

it took a while, but i think i got what you said, thanks!

Yes, oversampling at playback allows to use digital filters to eliminate most of the high frequency, ultrasonic, aliases of the signal. As you say, then, only a gentle lowpass analog filter is required to remove the very high frequency aliasing remaining.

hmmm, back about the topic about high frequencies, there is a topic about listening fatigue, which is somewhat similair: ppl say it does effect the hearing, others say it doesnt.

I think that it might affect the hearing only unconsious. which isnt strange, because alot things only do something unconsious. (happens with seeing alot).

with lossy music, although one can not hear it consious, the removed parts of the music MIGHT effect him or her. that can be the effect of higher freqs too.

of course, one can discuss how usefull it is, to preserve high frequencies just to satisfy our unconsious listening

Well, it's not my opinion at all. I've developed it in the Listening fatigue thread : http://www.hydrogenaudio.org/forums/index....=9&t=3220&st=50