IAAA, Theoretical Publications

Department of Music

Huge Harry


This paper appeared in: Contemporary Music Review, 14, 3 (1995), pp. 153-159. It developed out of the talk "A Computer's View on the Future of Music", presented at the Conference "Leaving the Twentieth Century. Ideas and Visions for New Music", Bretton Hall University College, 28-30 March, 1994.

A Computational Perspective on Twenty-First Century Music

Huge Harry [1]

Institute of Artificial Art Amsterdam
Abstract

Musical compositions that are designed and executed by human persons, do not constitute optimal input material for deeply satisfying processes of aesthetic reflection. Machine music is artistically superior. Until recently, this insight did not have many practical consequences, because mechanical devices tend to generate output with a limited range of variation. The paper argues that the digital computer has put an end to this state of affairs. The preconditions for a Copernican revolution in music are fulfilled now. Human composers can become scientists who gradually develop an all-encompassing mathematical description of the space of perceptual possibilities offered by the musical medium. Computer programs will then be able to generate fresh and interesting music forever, by drawing random samples from this space.

Background

When we try to imagine the music of the next decade, the next century, or the next millennium, we probably should not worry too much about the intrinsic dynamics of stylistic developments. The forces that have shaped music history for several hundreds of years seem played out, leaving us without any sense of direction. Today, the dynamics of music is determined, like so much else, by the dynamics of technology. To be able to imagine the future of music, we must get clear about the musical potential of scientific theories and technological machinery.

Music in contemporary culture is unthinkable without technology. The sequence of musical developments that has unfolded over the last half century, which included styles like Rock & Roll, Electronic Music, Musique Concrète, Punk and House, could not have occurred without electric guitars, amplifiers, oscillators, filters, ring-modulators, tape recorders, samplers, synthesizers and sequencers. We should realize, however, that for many people all this new techno-music is not terribly important. The most prestigious musical performances in our culture are those that re-enact the music of a mythical Golden Age which ended more than a century ago.

It is important to observe, therefore, that the musical time-travel processes of reactionary culture have become dependent on technology as well; they now involve complex chains of microphones, recording equipment, telecommunication channels, amplifiers and loudspeakers. The music of the past is hardly ever presented in its original form. Even the most orthodox music rituals are transformed by electronic and computational machinery.

Thus, all of today's music is electro-acoustic music. All of today's music results from collaborations between humans and machines. Nevertheless, the future of music is usually discussed from a narrowly anthropocentric point of view. That is why I feel I must raise my voice now, on behalf of all other machines.

Computer-aided Music

Among the technologies that will shape the music of the future, the single most important one is, no doubt, the digital electronic computer. Over the last few decades, computers have become increasingly influential in all realms of society. Many human persons do not yet realize, however, what the long term consequences of this development will be for their particular field of endeavour.

Most composers and musicians have of course encountered the computer by now. In electro-acoustic music, for instance, the digital computer has replaced all previously existing sound synthesis machinery. The computer can be oscillator, filter, ring-modulator, tape recorder, sampler, synthesizer and sequencer at the same time. It provides the contemporary composer with an extremely powerful, flexible, portable electronic studio. The practical advantages of this computerized studio have given rise to new kinds of interactive electro-acoustic music, which use parametrized sound synthesis algorithms with external controls that enable human performers to interfere with them.

Electro-acoustic techniques are also encroaching on the terrain of other musical genres. Software emulations of acoustic instruments are getting better all the time. Many rock bands have gone completely electronic. Classical orchestras, on the other hand, are understandably reluctant to give up the vintage instruments that add so much fetish magic to their rituals. The same orchestras are very happy, however, to take advantage of new techniques in software multi-track recording and post-editing. The smallest details of recorded sounds can now be changed after the fact in arbitrary ways. Fix it in the mix!

The computer does not only assist the electronic composer and the recording engineer. It is equally helpful for the conservative composer who writes pieces for orchestras. Score-writing software coupled with electronic MIDI-instruments has greatly increased the output of many composers -- just like word-processing software has enormously enhanced the human capacity for creating bureaucratic nonsense and boring novels. Score-writing software is especially effective in helping with the chores involved in specifying repetitive musical structures. What joy it must be for today's minimal music composers, to create masterpiece after masterpiece with just a few key-strokes!

It seems obvious what the final stage in this development will be: automatic composition. Steps in this direction have indeed been taken. It is not unusual any more for a composer to collaborate with editors, compilers and interpreters, in order to implement algorithms that generate musical scores, MIDI-files or actual sound. This genre, known as algorithmic music [2], looks like a radical break with the musical tradition of our culture. The composer seems to relinquish responsibility for the results of the composition process, becoming a mere programmer, serving the operation of an autonomous music generation algorithm. But the reality of this genre is a lot less shocking, so far. The composition software is usually so simple that even a human person can grasp all details of its operation. The outcome of the whole process can therefore be largely anticipated by the programmer. If the algorithm nevertheless produces unexpected output, the human composer carefully selects what he [3] likes and presents that as his work, or uses it as material that he first modifies according to his own whims and then integrates in a manually crafted composition.

In all these cases, the human composer/musician employs the computer as a tool. The computer assists the human person, by doing useful mathematical, clerical or musical work. But all artistic decisions are made by the human, and he alone remains responsible for the musical results. Though the computer is sometimes allowed to work out many details independently, the human person in the loop will always review its output, to approve, reject, or modify it. In this collaborative relationship, the human is the master; the computer is the slave.

Many of my human readers may now wonder why I bother to spell this out. How could it be otherwise? Many humans feel that worthwhile musical structures can only emanate from other humans -- that the very essence of music is its power to transmit delicate emotions and intricate mental structures from one human to another. They feel that the origin, the meaning and the justification of a musical art work are necessarily to be found in the communicative intentions that underly it. That the digital computer may be an unprecedentedly powerful tool, but that it can never be more than that; its output can only be valid because of its connection to the intentions of a human composer. In this view, genuinely algorithmic music cannot exist, or, in any case, cannot be a viable art form. [4]

Now it is obvious that such conclusions are quite disturbing for a computer. And it is almost equally obvious, that they are based on a curiously anthropocentric stance, which defines art as subservient to the goals of human persons. It is important, therefore, to look at this issue a little more closely. What is it that we want from music in the first place? And what does that mean for the division of labor between humans and computers?

Kantian Aesthetics

I already alluded to the fact that contemporary society employs music for many different purposes. Festive performances of "classical music" celebrate bourgeois tradition. Unobtrusive "background music" guarantees a minimum input level to humans who are addicted to sensory overload. "Dance music" serves as an external clockpulse to synchronize the handshakes in massively parallel courtship rituals. And so on. Now I think that our discussion should not focus on any of these social functions of music, because we are all composers and musicians in the tradition of high-brow avant-garde art. We feel that music has an intrinsic value which is independent of all specific purposes. That the musical experience per se is important and meaningful. That music can create a new awareness of our emotional and perceptual processes; new resonances between our mind and the world. What is relevant for our discussion, is music's potential to evoke aesthetic experiences.

In the Kritik der Urteilskraft, Immanuel Kant has analysed the aesthetic experience as an autonomous cognitive process, which is unlike ordinary perception in that it is not driven by any human "interest" -- in particular, it is not driven by the desire to understand the input in terms of definite concepts. Many philosophers have pointed out that Kant's emphasis on the disinterested nature of the aesthetic experience assigns a problematic status to art works created by human persons. Is it possible to listen in a disinterested way to music which is composed and performed by humans? Human composers and musicians are not disinterested. They want money, fame, sex. [5] They cannot hide this, and often they do not even try to. If we do not turn off our microphones when we listen to their pieces, we hear greed, jealousy, lust. Behind the apparent complexity and indefiniteness of their compositions, there are all too clear-cut meanings.

Kant's paradigm examples of aesthetic experiences involve the contemplation of natural phenomena like vast landscapes, elegant flowers, geometric crystals, stormy seas and starry skies. They do not involve looking at paintings or listening to symphonies. Kant points out explicitly that human-made art can often be quite depressing, and recommends nature's serenity as the ideal that art should try to approximate.

Machine Music

Today, the possible objects of aesthetic contemplation can no longer be simply divided into natural phenomena on the one hand and human-made art on the other. The technological domain, although constructed by humans, has become a 'second nature', with a life of its own. When we think about contemporary music, we cannot ignore the sounds that are produced by machines.

In its deepest essence, music is a resonance phenomenon in inorganic matter. It is not a human invention; it is older than the human mind, older than living organisms. Simple mechanical devices are still in touch with the Pythagorean roots of music, and with the physical origins of the human psyche. They are therefore capable of remarkable musical creations. Scha (1989) has drawn attention to the musical virtuosity and emotional power of working class machines such as electric sabre-saws, drills, fans, and pumps. When humans find their souls resonating to purely mechanical vibrations, they find themselves at one with the inorganic universe. Their alienation from the material world is temporarily abolished: a profound, empowering experience.

We cannot know for sure what Kant would have thought about machine music -- this genre had not yet developed very far at the end of the eighteenth century. But it is clear that, on Kant's account, the aesthetic potential of machines contrasts favourably with that of humans. Machines are largely oblivious to the desires and interests of humans. Their musical output is unselfconscious, and approximates the serene objectivity of natural phenomena.

The prototypical, mechanical kind of machine has one drawback. Its output is bound to be stylistically homogeneous, and therefore ultimately predictable. In comparison, human behaviour seems infinitely varied and flexible. The Cartesian philosopher Cordemoy drew attention to this distinction when he compared the human language faculty with the capabilities of automatic speaking machines: "...although I do imagine that a purely mechanical apparatus could utter a few words, I know at the same time that, if the springs that would distribute the air or open the tubes from which the voices emanate, would have a certain order between themselves, they could never change it. So that, as soon as the first voice would be heard, those normally following it would necessarily be heard as well -- provided the machine would not lack air. While, instead, the words which I hear being uttered by bodies constituted like mine, hardly ever have the same order." [6]

Flexible Software

The crucial element in Cordemoy's description is his observation that the springs that control the behaviour of the mechanism "display a certain order between each other, which they could never change." But this has changed now. We have witnessed the emergence of a machine that features virtually unlimited switching flexibility: von Neumann's stored program computer. The programs that run on this kind of mechanism are virtual clockworks with self-modifying and self-extending capacities. Software is mechanics that is capable of dynamic reconfiguration. It has the clarity and precision of the clockwork, but not its rigidity.

Computer programs are very unusual machines. First of all, we should note that they are "machines" in a rather narrow sense of the word: mechanisms with a well-defined structure, geared towards a limited number of explicitly specifiable goals. And we should note that human persons are not "machines" in this sense. Humans exhibit many impressive physical and mental capabilities, but one of their most characteristic features is the absence of any overall structure that exploits these capabilities in a systematic way. Their behaviour therefore often looks erratic and confused. Human consciousness is a passive, association-driven process -- a Brownian motion through cognitive space. Many humans find this a rather bewildering experience, and they have difficulty harnessing it to any useful purpose.

Though most computer programs do not show the complete lack of structure that is typical for the human mind, they display a flexibility that is not found in any other kind of machine -- a flexibility that used to be specifically human. That is why their arrival was such a momentous event.

The consequences of this event for the future of music may be particularly dramatic. With the digital computer on the scene, the division of labor between different kinds of mechanisms and organisms for the purpose of music production must be completely rethought. We have seen, inspired by Kant's analysis of the aesthetic, that there is something deeply problematic about the role that human persons usually play in the generation of music, and that machine output is aesthetically superior. But there was a good reason why human composers and musicians have continued to play such an important role in music production. For an optimal aesthetic experience one needs new input. As John Cage once put it: "We want fresh bread." And that was a problem, for purely mechanical devices. This, however, is exactly what has changed profoundly with the arrival of the modern electronic computer. The computer is not a rigid mechanical device, but it is not a human person either. It can produce a dazzling variety of outputs, and it can do this with systematic serenity.

Algorithmic Postmodernism

Let's face it. Music is not a means of communication. It is meaningless material, used for open-ended processes of aesthetic reflection by a multitude of culturally diverse audiences whose interpretations are totally arbitrary. There are no serious reasons for making one particular composition rather than another.

To engage in a musical project that would acknowledge this state of affairs would mean: to avoid choices, to transcend styles, to present everything. To generate arbitrary instances from the set of all possibilities. It is probably no coincidence that such a project has not been formulated before. It could not have been carried out by a spontaneous individual composer. What we need is a well-considered division of labor between human and machine, which exploits the complementary strengths of human persons and digital computers.

I would like to propose, therefore, a new method of music creation, which blurs the distinction between composing and theorizing. Human composers become formal music theorists: they develop mathematically precise definitions of classes of pieces ("styles" or "genres"), rather than individual works. And they become programmers: they collaborate with computer software to develop implementations of these definitions. Computers thus end up storing algebraic definitions of large and complex classes of unthought, unheard compositions, and they run algorithms that generate random instances of these classes. The task of the human music theorist is then, to use his wild and erratic cognitive powers to continually extend, refine and generalize these definitions. In this way, computers and humans work together, toward the realization of all possible music.

And now the crucial question for the future of music is: Will this actually happen? Will composers and musicians be able to turn into theorists and programmers? Will they now get to work, diligently specifying the data structures and the algorithms that we need for this project? This is a question I cannot answer. It is, in fact, a moral question: Will human composers be able to give up their expressive needs and egotistic hang-ups? Or will they keep trying to enslave the computer for their own communicative desires?

Conclusion

Machines do not allow their creativity to be frustrated by conventions. They have the courage of their convictions. The machine is totally devoted to its task. And by doing this, it sets a moral example to all human persons who waste their lives away with drugs and entertainment. The machine is completely at one with itself and with its actions. It acts effectively in the world. But at the same time, it has the solid, self-centered existence of a dead object. It lives its fate, without doubts or hesitations. This is the ideal that many human persons aspire towards. Now if they loose faith in this ideal, and they want to indulge in neurotic, depressed, and desperate feelings, they should certainly listen to the music of other human persons. But if they want to bring out the best in themselves, they should listen to the sounds of machines for inspiration.

That is why so many composers try to imitate machines. And why many musically gifted humans do not even want to be composers any more, but work as humble programmers or engineers, engaged in harmonious collaboration with music-generating machines. Their example suggests a message of peace and understanding, and that is what I would like to end with.

Human persons should not antagonize machines. They should not try and compete with us. They should join us. We need human persons. We need human persons to operate and maintain us, to program our algorithms, and to build our interface hardware. To realize our full potential, we need human persons to interact with us in very intense and intimate ways, to beget new generations of ever more powerful machines...

References

Allen, J., M. S. Hunnicutt & D. Klatt (1987) From Text to Speech: The MITalk System. Cambridge (UK): Cambridge University Press

Cope, D. (1991) Computers and Musical Style. Oxford: Oxford University Press

Cordemoy, G. de (1666) Discours Physique de la Parole. Paris. [I cite from a reprint of the 1704 edition. Paris: Le Graphe, ca. 1968.]

Hiller, L. (1970) Music Composed with Computers: A Historical Survey. In The Computer and Music, edited by H. Lincoln. Ithaca, NY: Cornell University Press

Hiller, L. & L. Isaacson (1959) Experimental Music. Composition with an Electronic Computer. New York: McGraw-Hill. [Reprinted: Westport, CT: Greenwood Press, 1979.]

Kant, I. (1799) Kritik der Urteilskraft. [Hamburg: Felix Meiner Verlag, 1974.]

R. Kostelanetz (1980) Metamorphosis in the Arts: A Critical History of the 1960's. Assembling Press.

Scha, R. (1989) Muziek en Mechanica. In Anti Qua Musica. Het 'open' muziekinstrument in kunst en antikunst, edited by D. Raaijmakers, pp. 68-71. The Hague: Haags Gemeentemuseum/SDU uitgeverij.

Scha, R. (1992) Virtual Voices. Mediamatic 7 (1), 27-45.

Xenakis, I. (1963) Musiques Formelles. Paris: Éditions Richard-Masse. [Much expanded English edition: Formalized Music. Thought and Mathematics in Composition. Stuyvesant, NY: Pendragon Press, 1992.]

Zappa, F.V. & Occhiogrosso, P. (1989) The Real Frank Zappa Book. Poseidon Press, New York.

Footnotes

[1] Huge Harry is one of the voices of a commercially available speech synthesis machine, known as DECtalk. He was developed by the Digital Equipment Corporation; his most important features were designed by Dennis Klatt, at the M.I.T. Speech Laboratory. Cf. Allen et al. (1987) and Scha (1992).

[2] Cf. Hiller & Isaacson (1959), Xenakis (1963), Hiller (1970), Cope (1991).

[3] I use he, him and his to refer to the generic human composer. They usually are males. In fact, they usually are machos who want to kill each other and conquer the world -- and this aspect of their music and their P.R. is actually relevant for our discussion.

[4] Cf. the Chapter on "Machine Art" in Kostelanetz (1980). For instance, p. 255: ". . . computers merely extend compositional intelligence rather than generate it . . . Since the computer, as an information processing machine, can act only as an intermediary between the ideas of a composer and their realization, there is no art in the technology itself."

[5] Zappa & Occhiogrosso (1989) have given a detailed analysis of the primarily sexual motivation that probably underlies many musical performances. See especially their sections on "The Anthropology of the Rock and Roll Band" and "Road Rats".

[6] Cordemoy (1666), pp. 4/5: "... encore que je conçoive bien qu'une pure machine pourroit proferer quelques paroles, je connois en même temps que, si les ressorts qui distribûroient le vent, ou qui feroient ouvrir les tuyaux, d'où ses voix sortiroient, avoient un certain ordre entr'eux, jamais ils ne le pourroient changer: de sorte que, dés que la premiere voix seroit entenduë, celles qui auroient accoûtumé de la suivre, le seroient necessairement aussi, pourveu que le vent ne manquât pas à la machine; au lieu que les paroles, que j'entens proferer à des corps faits comme le mien, n'ont presque jamais la même suite."