From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!sun-barr!olivea!uunet!haven.umd.edu!darwin.sura.net!jvnc.net!nuscc!hilbert!smoliar Thu Apr 30 15:23:00 EDT 1992
Article 5289 of comp.ai.philosophy:
Xref: newshub.ccs.yorku.ca comp.ai.philosophy:5289 rec.music.classical:10656
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!sun-barr!olivea!uunet!haven.umd.edu!darwin.sura.net!jvnc.net!nuscc!hilbert!smoliar
>From: smoliar@hilbert.iss.nus.sg (stephen smoliar)
Newsgroups: comp.ai.philosophy,rec.music.classical
Subject: Re: Intelligence, awareness, and esthetics
Message-ID: <1992Apr27.231026.13791@nuscc.nus.sg>
Date: 27 Apr 92 23:10:26 GMT
References: <VANCE.92Apr23105018@kyoto.speech.sri.com> <1992Apr23.235522.25154@nuscc.nus.sg> <1992Apr24.153412.15904@javelin.sim.es.com>
Sender: usenet@nuscc.nus.sg
Reply-To: smoliar@iss.nus.sg (stephen smoliar)
Organization: Institute of Systems Science, NUS, Singapore
Lines: 43

In article <1992Apr24.153412.15904@javelin.sim.es.com>
biesel@javelin.sim.es.com (Heiner Biesel) writes:
>Aside from the specific instances referred to by Stephen and others, it
>seems that music generally relies upon a subtle balance between the
>expected and known, and carefully structured and integrated novelty.

Take a look at Roger Schank's DYNAMIC MEMORY;  he generalizes this principle
from music to general comprehension.

>This is true both of music in general - viz the variuos classical forms -
>and of individual composers in particular. We know it's Mozart when we hear
>it. Why? Chordal progressions, etc. offer a clue, but are far from the whole
>story. Is it in principle possible to analyze the style of a composer to
>the point where new compositions indistinguishable from the original could
>be produced by a machine?

The first thing you should do is ask, "Is this something that people do?"  The
answer is, "All the time."  The standard pedagogical approach to "free
composition" is to give the student a task of imitation, generally involving
the use of given work as a model and "making it different."  I am not quite
sure how long this technique has been around, but I think it goes back to the
time before Mozart.  (I am again cross-posting to rec.music.classical in the
hope that some musicologist can help on this point.)  The only point I wish
to make is that this is a relatively well-defined human task which occasionally
leads to interesting results.

Can a machine take on the same task?  This has basically been the research
topic of David Cope for several years.  He has been working on software which
essentially takes a "model" composition as input and generates another one.
He has definitely progressed to the point where he can fool some of the people
some of the time, but I am not sure that is his objective.  On the other hand
what he may be demonstrating is that a few surface features based on archetypal
motifs may have more to do with recognizing the work of a given composer than
many of our more elevated principles about style and composition.  Cope's
article in the July 1991 issue of COMPUTER is definitely worth the read,
and one should certainly try to listen to the recorded results on the
accompanying compact disc.  (After that, you can read my review of his
work in the next issue of IEEE EXPERT!)
-- 
Stephen W. Smoliar; Institute of Systems Science
National University of Singapore; Heng Mui Keng Terrace
Kent Ridge, SINGAPORE 0511
Internet:  smoliar@iss.nus.sg


