I began Experiments in Musical Intelligence in 1981 as an attempt to create new instances of music in my style. With a lack of quantifiable definitions of style, I concentrated on the commonalties in the works of certain composers, commonalties I call signatures. By 1987 Experiments in Musical Intelligence had produced works (arguably) in the styles of Bach and Mozart, among others. Further experimentation allowed for more extensive output both in terms of work length and complexity as well as stylistic diversity Experiments in Musical Intelligence subsequently produced new works in the styles of composers as contrasting as Stravinsky, Palestrina and Joplin. These works have been discussed and, in part, reproduced in my books Computers and Musical Style (1991), Experiments in Musical Intelligence (1996), and The Algorithmic Composer (2000) published by A-R Editions, Madison, Wisconsin and Virtual Music soon to be released from MIT Press. Many Experiments in Musical Intelligence works have been recorded on Centaur Records Bach by Design, Classical Music Composed by Computer and Virtual Mozart (music recorded four years ago at another orchestral concert collaboration of the Baroque Festival and April in Santa Cruz).

I am often asked about Experiments in Musical Intelligence’s value beyond its novel use for composing works in various styles. Certainly the program has uncovered various new approaches to better understanding musical style. Such revelations are not trivial as what exactly constitutes the differences between not only Mozart and Beethoven, but between early Mozart and late Mozart, has not been previously well researched. As well, most current computer programs capable of composing music create quite novel and interesting results. Rarely, however, does the music they produce meet an immediate need for composers. Experiments in Musical Intelligence, on the other hand, serves a very practical purpose—supplying relevant and usable music integral to existing in-progress works. Once honed, it would seem that programs like Experiments in Musical Intelligence should prosper as composers can request and receive music in their own style whenever desired. I believe that twenty-first century composers of all types will routinely collaborate with such computer programs.

The Experiments in Musical Intelligence works performed here are world premieres. They also represent ‘firsts’ in that no other machine-composed concerti or suites in the style of Bach have ever been created or performed previously.

The Experiments in Musical Intelligence Bach Suite for solo violoncello (2000) follows the forms of Bach’s six cello suites quite closely. Each movement has a repeated binary form with tempos and general rhythmic figurations quite similar to, but not duplicating, those of the originals.

The Experiments in Musical Intelligence Bach Brandenburg Concerto (2000) is based on the six Bach Brandenburg Concerti as well as certain similar movements from Bach’s orchestral suites (used to increase the size of the database). The form follows the three-movement structures of most of its models. The first movement is somewhat unusual in that none of the actual Brandenburg Concerti begin in minor—this, no doubt an influence of either another movement in the concerti or a movement of the orchestral suites.

The Experiments in Musical Intelligence Bach Harpsichord Concerto (2000) also has three movements with the first clearly using Bach’s D-minor concerto as a model, even though the music differs in substantial ways. The second movement resembles many Bach keyboard concerti second movements, though none so much as the second movement of the F-minor concerto and the Air from the third orchestral suite, though again the differences are significant. This concerto is dedicated to tonight’s soloist Linda Burman-Hall.

The databases of Bach’s original music used by Experiments in Musical Intelligence were provided by the Stanford Center for Computer Assisted Research in the Humanities (CCARH) to whom I am very grateful. I also wish to thank Linda Burman-Hall, Nicole Paiement, and Leta Miller, my colleagues in the UCSC Focused Research Activity in Performance Practice, and the UCSC Committee on Research, for grant support.