Klarinet Archive - Posting 000718.txt from 1997/11

From: Jonathan Cohler <cohler@-----.net>
Subj: Re: Nyquist and analog
Date: Thu, 20 Nov 1997 11:18:12 -0500

>Can anybody join in????
>On the face of things, human ears don't perform Fourier analysis.

To the contrary. That is in fact a very good way of visualizing how the
human ear works (in particular the cochlear nerve).

Briefly, the cochlear nerve is a long wound up nerve (somewhat resembling a
snail, I'm told) inside the ear. The nerve has "hair cells" which are
excited by sound, and it is these that send the signals to the brain which
interprets the sound.

If one were to unwind the cochlear nerve in a straight line, and lay it on
an immaginary X-axis, then it is precisely the position along this X-axis
that determines the associated frequency of the corresponding hair cells.

In other words, the cochlear nerve is essentially a bioligical fourier
analyzer.

> Some of
>what has been said seems to suggest that they do, with this stuff about
>hearing or not hearing "components". I thought it was accepted that even
>what would be inaudibly high *components* of an audio signal (meaning, if
>anyone's forgotten, a *musical sound* in the first place) contribute
>additively to what is heard.

The brain hears upper harmonics as colorations of tone. However, one can
play pure sine waves up to 20KHz and individuals with excellent hearing can
just barely notice them up there.

The reason the brain doesn't notice the upper components of a complex tone
as seperate tones is because their amplitude is much smaller than the
amplitude of the fundamental. Furthermore, the ear's sensitivity falls off
at high frequencies.

To my knowledge, however, noone has ever done a double-blind study to show
that frequency content above 20KHz has any noticeable effect on tone color.
Which, of course, makes sense, because, for example, if we can't notice a
25KHz tone when it is played by itself at a large amplitude in a totally
silent environment, then why should we be able to notice it suddenly when
it is played together with some other sub-20KHz tone?

>
> Even if all the components below 20 KHz can
>be faithfully reproduced, then the original sound is compromised - just as
>it is with analogue recording, for different reasons.

The waveform may be compromised (if it had components above 20KHz), but the
sound (that which we hear) is not.

In analog LP recordings, many things are compromised quite dramatically and
audibly including dynamic range, frequency response, noise from the medium
etc....

Some people prefer the compromised sound, however. In essence, they do not
want to hear the actual sound as it was originally performed. They prefer
the distortions introduced by the playback system.

----------------------
Jonathan Cohler
cohler@-----.net