THE ELECTRONIC COMPUTER TAKES UP MUSIC LESSONS

by

Ivor Darreg - 6 June 1957

In the June, 1957 issue of Radio-Electronics magazine, beginning on page 36, is an article by Dr. martin L. Klein entitled "Syncopation in Automation." At the top of the first page is reproduced the "lead sheet" of a popular tune, Push Button Bertha, with words by Jack Owens and music by a Datatron computer. The article says that there was difficulty about obtaining a copyright for the tune, as the Copyright Office never before had been confronted with a tune "written by a machine."

After an explanation of how the "memory units" (internal information storage devices) work, and an account of the flipflop circuits which do the arithmetic in computers, there is a brief explanation of the strict rules governing the writing of popular-song tunes. It is really amazing that this has not been pointed out before--the average jukebox hits are so much of a pattern that, in Dr. Klein's words, "it sounds as if the same piece of music had been written over and over."

Rules deduced from a statistical examination of 100 top popular tunes over a one-year period were "coded" into a form that could be used to program the Datatron computer. The machine then applied those rules to a set of random digits (such as would be obtained,for example, by a chance selection of numerals from pages of the telephone directory) and printed out on its electric typewriter a jumble of letters, figures, and symbols corresponding to the notes of a popular song melody. By this means, 4000 tunes per hour can be produced, which at first blush would seem to put all the songwriters out of business!

Somehow, this writer doesn't think so. Computers, although they can do more arithmetic faster than anybody, are not really "giant brains." Their memory capacity is limited, as much by economic considerations as anything else, and they lack creative imagination. A computer can only do what it has been instructed to do--and explicitly instructed. But this does not lessen the value to composers, arrangers, and others of a new and highly useful tool.

"Genius is 10% inspiration and 90% perspiration." Very well, we have here an apparatus that may well take care of most of that 90% perspiration. For example, given a short theme, the four or five notes may be fed into the machine, pertinent instructions punched into cards or tape, and in a few seconds one would hear the theme played at various speeds, in various tone-qualities, in various keys, and even in counterpoint against differently-permuted versions of itself.

SImultaneously with the playing of the different versions of the theme, a recording would be made. Form this recording the composer would select the usable forms, erasing the unsuitable ones, and on the machine's keyboard, punch out the proper code combinations for various harmonizations and rearrangements of the theme or themes. What this means is, that all during the composing process (or arranging process, in the case of an arranger), the themes and their developments would be heard just as they were going to sound in the finished piece, not in piano tones nor merely imagined from the written score.

In the terminology of information theory, music is highly redundant. That is, nearly all music is highly repetitious, and from a relatively very small amount of new information (i.e., just a few freshly-invented melodic and rhythmic patterns), very long pieces can be elaborated. There are both aesthetic and psychological reasons why this is so, for the further information on this we will refer you to the standard books on the subject. What matters here is that in the present methods of composition and arranging, there is too much purely mechanical clerical work involved. What, then, could be more logical (in several senses of the term 'logical'!) than to have an electronic computer do this clerical work? There is nothing more tedious to a composer than to have to write down the same measures several times over, with small variations.

Now, some readers are going to object that all this mathematics, mechanics and electronic engineering will make the music sound mechanical. Our reply would be: pianos, organs, and orchestral instruments are all obviously mechanisms, whereas electricity works invisibly; music already has a mathematical basis and a physical basis, so why not take advantage of recent discoveries in those fields? Music is a form of communication and is therefore just as adaptable to data-processing machinery as any other kind of communication is. Composers will still have to possess musical talent; there is no substitute for originality.