SCHWANAUER, STEPHAN / LEVITT,DAVID
The editors' premise in assembling these twenty-three essays is that since it is possible to model human thought, particularly advanced logic and artificial intelligence, then it should also be possible to model creative thought, and in particular, musically creative thought. Machine Models of Music is not entirely about getting machines to be creative; it is also about understanding the human mind by modeling its approximate behavior algorithmically. Much of it is very technical, e.g., author Charles Ames discovery that "algorithmic methods of composition could be much more powerful than had previously been demonstrated..." as published in the 1981 International Computer Music Conference. We are also treated to somewhat more accessible papers, such as James Anderson Moorer's article, "Music and Computer Composition" which views music modeling and speech modeling as closely-related disciplines. For a fun challenge, see the "Musical Dice Game" written by Wolfgang Amadeus Mozart, which shows how to compose waltzes with two dice. I took particular interest in the editor's own paper, "A Learning Machine for Tonal Composition," by Stephan Schwanauer. Schwanauer (aka, "The Count of Palo Alto") developed The Music Understanding System Evolver (MUSE) as his Ph.D. thesis at Yale in 1986 with the cooperation of Yale's Music and Computer Science Departments.
At its writing, this book was a pioneering effort in harmonizing the art of music with the science of music. In the process it took a giant step toward its fundamental objective of understanding the mechanics of human thought.