00100	COMMENT ⊗   VALID 00004 PAGES
00200	C REC  PAGE   DESCRIPTION
00300	C00001 00001
00400	C00002 00002	
00500	C00008 00003	Audio signals are generated by a PDP-10 computer which has a 4-channel D-A
00600	C00009 00004
00700	C00011 ENDMK
00800	C⊗;
     

00100			NSF Proposal
00200	
00300		Computer Simulation of Natural Auditory Signals
00400			and Reverberant Spaces I.	Introduction
00500	
00600	In  recent  years  there  have  been significant advances in computer
00700	simulation and analysis of complex  auditory  signals.   On  the  one
00800	hand,  the  use  of  the  computer has demonstrated the extraordinary
00900	complexity of natural sounds and consequent difficulty in simulation,
01000	and   on   the   other,  the  computer  has  provided  the  means  of
01100	implementation of powerful synthesis and analytical techniques.
01200	
01300	At the  Stanford  Artificial  Intelligence  Laboratory  programs  and
01400	techniques  have  been  developed  for the simulation of moving sound
01500	sources in reverberant spaces and for the analysis and  synthesis  of
01600	complex   signals,   some   of   which   are  startlingly  simple  in
01700	implementation and novel in conception.  While the progress  of  this
01800	research   has   certainly   been  enhanced  by  the  high  level  of
01900	technological expertise at the Artificial Intelligence Laboratory, it
02000	has  been  equally  dependent  upon the cross-disciplinary skills and
02100	insights of the researchers themselves.  It is  this  latter  aspect,
02200	science/music,  which has given our research its particular character
02300	and which circumscribes our interests.
02400	
02500	The Perception of Timbre
02600	
02700	Music is based upon,  and  indeed  is  dependent  upon,  signals  and
02800	environments  which  are infinately more complex than those which are
02900	commonly used in psycho-acoustical research.  Yet, it  is  well-known
03000	that  trained  listeners  are  capable of making extraordinarily fine
03100	distinctions between complex signals  in  complex  environments.   In
03200	some  of  the  dimensions of acoustical phenomena, such as pitch, the
03300	nature of the distinctions is fairly  well  understood.  However,  in
03400	another  dimension, timbre (sometimes called tone-color), there is no
03500	coherent theory to explain either the physical characteristics  of  a
03600	sound  to which the ear responds or the mode of organization effected
03700	in perceptual processing.
03800	
03900		Methodology- We have discovered a technique for synthesis  by
04000	means  of  FM  [reference]  which  is  capable  of  producing for the
04100	listener a strong impression of a variety of natural music-instrument
04200	tones.   The spectrum produced by this technique, however, differs in
04300	major and conspicuous ways from the spectrum of a natural tone  which
04400	produces  a similar impression.  This discrepancy has led us to begin
04500	a rigorous investigation  of  those  physical  characteristics  of  a
04600	complex spectrum which are perceptually salient.
04700	
04800	In  order  to  develop  a  methodology,  another  powerful  synthesis
04900	technique was created which allows  temporal  control  of  phase  and
05000	ampliotude for n partials. It is capable of synthesizing a tone which
05100	includes all of, or a simplification of, the spectral detail  derived
05200	from  a  computer  analysis  of  a  natural  tone,  therby allowing a
05300	systematic reduction in data of a  spectrum  and  comparison  of  the
05400	audible waveform with the unaltered original.
05500	
05600	The  two synthesis techniques are applied to the same natural tone of
05700	interest. The FM synthesis data is increased to reach  an  acceptable
05800	simulation,  while  the  additive  synthesis  data  is  decreased  to
05900	maintain an acceptable simulation.   Examination  of  the  difference
06000	between  the  resulting  spectra provides the information for another
06100	synthesis pass.  The  results  of  this  method  should  confirm  our
06200	success  at  having isolated those physical features of spectra which
06300	correlate to the perception of timbre.
06400	
06500		Testing and Scaling-
     

00100	Audio signals are generated by a PDP-10 computer which has a 4-channel D-A
00200	converter for analog conversion of the sampled waveform.
00300	The analysis is accomplished by means of an A-D converter which converts
00400	audio input into a sampled wave which is stored on a disk for processing.
00500	The programs forthe synthesis of wave-forms are based on the acoustical
00600	compiler developed by M.V. Mathews at Bell Telephone Laboratories
00700	adapted to the time-sharing, interactive capabilities of the PDP-10 system,
00800	while the wave analysis programs were written by J.A. Moorer of Stanford.
00900	
01000