[net.audio] Processing digital audio

yraurb@hou2g.UUCP (G.DECKER) (04/05/84)

I was browsing through the latest Phillips Technical Review and came across
an article on digital audio circuits and listening tests.  They have develop-
ed a computer system that can simulate different digital circuit configur-
ations for signal processing and they can do this in real time.  They im-
plemented Fletcher-Munson curves and reverberation.  It occurred to me that
a mortee correct implementation of Fletcher -Munson curves would track the 
instantaneous level of the program since the "correction" should take into
account the absolute l;evels as well as the volume difference between live
and reproduced.  Anyway, they concluded that word lengths of over 22 bits
are necessary to maintain 16 bit quality from input to output when doing 
this kind of processingsignal processing for delay lines of reverberation.r reverberation. 

spoo@utcsrgv.UUCP (Suk Lee) (04/08/84)

It's clear that the Fletcher-Munson curves should *not* track
the instantaneous signal level (as described by DECKER).  When
we listen to live music, nothing is done to correct the 
frequency response abnormalities in the auditory system, so why
should we make an after-the-fact correction?  I suspect that if
such an instantaneous correction were applied, it would be very
disagreeable--the intruments would appear to have more and more
bass as they get quieter, quite an unnatural phenomenon.



-- 

From the pooped paws of:
Suk Lee
..!{decvax,linus,allegra,ihnp4}!utcsrgv!spoo