Digital audio listening test

[Martin Czech]

The sound pressure is transformed from the ear drum via anvil, hammer
and -the third I've forgotten - bones to the cochlea window. This is
an impedance transformation, strong movement , low pressure into less
displacement, high pressure, because the cochlea is filled with pretty
much inccompressible water.

Assuming that the cochlea is linear (it is not) there is still
conciderable distortion due to nonlinear action of the three tiny bones.

So, even if test persons are not able to hear low pressure ideal sine
waves higher then 17 kHz (this is my personal limit as I have tested
it recently) via good headphones, it can very well be that they hear
intermodulation products of theoretically inaudible frequencys due to
mix down.

The sine/square experiment leeds to the following:

The sine has (after processing) some distortion artefacts (which should
be inaudible, cause to high).  The square already has partials, the sum
and difference frequencys can be as low as the fundamental.

So even with my simple assumption above the ear should notice some
change in the fundamental energy, and this is audible.
At frequencys that high test persons (and me) can not really
tell what the difference is, but they still notice.

Things should be even wilder if two sine waves are intermodulated in
the ear, preferably with non integer frequency ratio.
Anyone who has listened to two or more not-so-good flute (recorder)
players will know what I mean. Ugly side products that give humming
noises in your ear and make you realyl mad.

So , brickwall limiting takes something audible away, no doubt.
The question is if I like these mixed products of the real world
or prefer the "cleaner" recording.


m.c.