"Don Pearce" wrote in message


JIm, there is another point you might want to consider,
which is that Shannon's limit applies only to data
transmitted over the defined path. But music in
particular is not like that. When we listen to familiar
music it is already stored in our brain and all that is
necessary to trigger it is a very small correlation from
the signal; it is a little like CDMA from this point of
view - a clean signal can be recovered from an apparently
impossible signal to noise ratio because the spreading
code is known at both ends of the link, and doesn't not
actually get transmitted and received.

IOW, the amount of data transmitted over the given data link at that time is
far less than the amount of correct data that is present at the receiver,
because the receiver has other transmissions in its memory that it uses to
develop the correct data that is perceived at that time.

So a familiar tune sunk well into the noise will be
heard, while random notes may be missed.

....depending on the reliability of the data that has been stored in memory
and the quality of the new data.

I know this is not what your article is about, but I
think it is relevant and you never know, may be worth a
research grant to quantify the effect.

This is a way to explain how people can hear what they say they hear when
they listen to vinyl. ;-)