We use cookies to enhance your experience on our website. By continuing to use our website, you are agreeing to our use of cookies. You can change your cookie settings at any time. Find out more


Music in the Late Twentieth Century


CHAPTER 10 Millennium's End
Richard Taruskin

The beginning of “computer music” can be specified with greater precision than perhaps any comparable event in the history of music. It took place in Summit, New Jersey, in 1957, when Max V. Mathews (b. 1926), an engineer at Bell Telephone Laboratories, produced computer-generated musical sounds with his “transducer,” an instrument he invented that could convert audio signals into digital information that could be stored or manipulated by a computer and then reconverted into audio signals. (Its initial purpose was to simulate and recognize speech so that some of the tasks that telephone operators performed could be automated.)

At first, what attracted the interest of engineers and, eventually, composers to the new medium was the classically modernist prospect of overcoming all limitations on creative sovereignty. “With the development of this equipment carried out at the Bell Telephone Laboratories, the composer will have the benefits of a notational system so precise that future generations will know exactly how the composer intended his music to sound,” Mathews wrote. “He will have at his command an ‘instrument’ which is itself directly involved in the creative process.” And even more grandly:

Man's music has always been acoustically limited by the instruments on which he plays. These are mechanisms which have physical restrictions. We have made sound and music directly from numbers, surmounting conventional limitations of instruments. Thus, the musical universe is now circumscribed only by man's perceptions and creativity.52

The new freedom came at a high price. The composing process for computer-assisted music was at first almost unbelievably cumbersome, and would remain so for a long time. Here is how Mathews, its inventor, described it in prospect:

Any sound can be described mathematically by a sequence of numbers. Our composer thus begins by determining what numbers specify the particular sounds in which he is interested. These numbers are then punched on IBM cards; the cards are fed into the computer and the digits recorded in the memory of the machine. The computer is thus able to generate limitless sounds, depending on the instructions given it by the composer. The latter, instead of writing the score in notes, programs his music by punching a second set of IBM cards, which when fed into the computer cause it to register on tape certain sounds from its vast storehouse.53

But here is how Kyle Gann, a composer, described it in retrospect:

From these early days to the late 1970s, computer music was made by punching Hollerith computer cards in stacks of maybe 3,000 for a few seconds’ worth of music, sending those cards out to a mainframe computer for processing, then having the resulting number-coded tape run through a digital-to-analogue converter to get actual sound. This generally meant punching your cards and waiting two weeks for them to come back—often only to find that some number error or miscalculation had torpedoed the desired results.54

The composers willing to pay the price, naturally, were the composers to whom the prospect of infinite control was most attractive. This chiefly meant Princeton and Columbia composers at the outset, and only partly because their universities were located within forty miles of Bell Labs. “Using a computer,” wrote Charles Dodge (b. 1942), who studied at both universities, “it is realistically possible for a composer to structure all elements of his composition (e.g., tempo, timbre, rate and shape of attack and decay, register, etc.) to the same degree as pitch and rhythm.”55 To “structure,” in those days and in those places, of course meant to serialize.

The earliest fruits of the Bell transducer program, presented on a commercial LP record called Music from Mathematics, consisted in the main of little experimental pieces by engineers in which the purpose was working the kinks out of the program. (The quotations given above from Mathews come from the sleeve notes to this recording.) Since the best way of testing the accuracy of one's results was to measure it against a known prototype, one of the bands on the record contained a sixteenth-century fantasia for three computer-simulated recorders, and another was the voice-synthesized rendition of “A Bicycle Built for Two” (“Daisy, Daisy”) that became famous when Stanley Kubrick put it to dramatic use in his futuristic movie 2001: A Space Odyssey.

But the record also contained two twelve-tone studies by David Lewin (1933–2003), later an eminent theorist and music analyst, who had graduated from Harvard in mathematics and from Princeton in composition, and whose computer pieces were attempts at serializing multiple “parameters” like duration and register. (But also featured was an unnotatable Noise Study by James Tenney (1934–2006), a disciple of Cage, in which complex sounds were allowed to modify one another in a manner that proved more indicative of things to come.) The first course in computer music technology was offered at Princeton in 1966 by Godfrey Winham as part of the university's newly approved Ph.D. program in composition. Among the earliest composers to use the computer regularly as a creative instrument was J. K. Randall (b. 1929), a Princeton professor whose earliest computer compositions, like Lewin's, sought serial control over every measurable aspect of the musical result.

Soon afterward a computer music-synthesis program was established at Columbia, Princeton's electronic music partner; it, too, was affiliated (through Vladimir Ussachevsky) with Bell Labs. Dodge, a widely noted young serialist then completing a doctorate in composition, was its star. At first, in keeping with his enthusiastic comment quoted above, he was attracted to the computer as a sort of performer—an instrument capable of coping (in the words of Kurt Stone, a celebrated music editor of the period) “with superhuman structural and interpretative complexities so typical of much of today's music.”56

Changes (1969–70), Dodge's first computer composition, resembled his earlier pieces for conventional instruments (like Folia, a very complicated chamber nonet commissioned by the Fromm Foundation and performed at Tanglewood in 1965), and used the new medium very much the way Milton Babbitt had used the big RCA synthesizer. The machine was programmed to store digitally and convert into an analog recording a fully written-out twelve-tone composition that live performers might also perform (provided they could be programmed with equal precision).

Dodge's next computer piece, though, was a turning point. He received a commission from Nonesuch Records (an adventurous company that had already had great success with Silver Apples of the Moon, a composition produced by Morton Subotnick directly on tape using an electronic synthesizer) to create a piece of computer music that could be marketed on disk. The work he produced, The Earth's Magnetic Field, was based on numbers derived from measurements that geophysicists at Columbia's Goddard Institute for Space Studies had taken of fluctuations in the earth's magnetism caused by the sun's radiation (the so-called “solar wind”) over the course of the year 1961. The measurements are averaged every three hours (for a yearly total of 2,920 readings) according to a scale known as the Kp index, which has twenty-eight values or degrees of magnitude (Fig. 10-7).

On the second side of the record, Dodge arbitrarily assigned the twenty-eight values to the tones of an equal-tempered chromatic scale covering two octaves plus a major third (twenty-eight semitones), and from other aspects of the geophysical data he derived some equally arbitrary but consistent rules for varying tempo, dynamics, and timbre. The result was an atonal and dissonant texture, not organized according to twelve-tone principles but similar to twelve-tone music in harmonic effect. That is, it was harmonically undifferentiated and undirected, and was therefore typical of the academic music of Dodge's generation. Nothing newsworthy about that.

The Elite Phase

fig. 10-6 Graph of sunspot activity used to control aspects of the composition in Charles Dodge's Earth's Magnetic Field.

The Elite Phase

fig. 10-7 Dodge, Earth's Magnetic Field, Kp readings, including the time interval January 1 to February 6, 1961.

On the first side of the disk, however, Dodge accepted a correlation between Kp measurements and musical pitches that had been previously worked out by one of the astrophysicists at Goddard, a musical amateur who mapped the measurements onto a diatonic scale that covered four octaves and a second. The result was a series of quirky, catchy tunes that listeners (perhaps with the kind of chemical assistance that prevailed among lovers of “alternative musics” in 1970) could imagine the sun “playing” on the terrestrial atmosphere. The record sold like hotcakes.

Dodge never returned to serial music. Instead, he became particularly interested in speech synthesis (the original Bell project) and its musical applications. A set of exercises called Speech Songs (1972), in which Dodge read a set of little poems by his friend Mark Strand several times into Mathews's transducer and then electronically resynthesized, modified, and mixed the sounds of his own readings, became a cult classic. “Laughter at new music concerts, especially in New York these days, is a rare thing,” the composer wrote in 1976, “and it has been a source of great pleasure to me to hear audiences respond with laughter to places in all four of the Speech Songs.”57 The second in the series plays on the cusp between the obviously synthesized and the possibly “real” sound of a voice.

In 1980 Dodge produced a hilarious sequel: Any Resemblance Is Purely Coincidental, in which the raw material was the recorded voice of the great operatic tenor Enrico Caruso. Against expectations, work with the computer had nudged him away from the serious but hermetic project that originally attracted him, and also away from musical idioms that could exist in notation apart from the new medium. At the same time, high technology conferred sufficient prestige to license (or excuse) an “accessible,” even humorous musical result that might again court the “lay” audience academic musicians had, it seemed, permanently forsworn. Computers, of all things, seemed to promise relaxation of the quixotic standards of difficulty by which composers in the academy had sought to justify their existence. Improbably enough, the most advanced technology was leading some of its elite practitioners toward a postmodernist posture.


(52) Max V. Mathews and Ben Deutschman, liner note to Music from Mathematics, Decca Records DL 79103 (ca. 1962).

(53) Ibid.

(54) Gann, American Music in the Twentieth Century, p. 266.

(55) Charles Dodge, liner note to Computer Music, Nonesuch Records H-71245 (ca. 1970).

(56) Kurt Stone, “Current Chronicle: Lenox, Mass.,” Musical Quarterly LI (1965): 690.

(57) Liner note to Synthesized Speech Music by Charles Dodge, Composers Recordings CRI SD 348 (1976).

Citation (MLA):
Richard Taruskin. "Chapter 10 Millennium's End." The Oxford History of Western Music. Oxford University Press. New York, USA. n.d. Web. 23 Apr. 2021. <https://www.oxfordwesternmusic.com/view/Volume5/actrade-9780195384857-div1-010010.xml>.
Citation (APA):
Taruskin, R. (n.d.). Chapter 10 Millennium's End. In Oxford University Press, Music in the Late Twentieth Century. New York, USA. Retrieved 23 Apr. 2021, from https://www.oxfordwesternmusic.com/view/Volume5/actrade-9780195384857-div1-010010.xml
Citation (Chicago):
Richard Taruskin. "Chapter 10 Millennium's End." In Music in the Late Twentieth Century, Oxford University Press. (New York, USA, n.d.). Retrieved 23 Apr. 2021, from https://www.oxfordwesternmusic.com/view/Volume5/actrade-9780195384857-div1-010010.xml