IAAA, Theoretical Publications

Department of Music

Huge Harry

DECtalk code of Huge Harry's lecture at the Conference "Leaving the Twentieth Century. Ideas and Visions for New Music." March 28-30, 1994. Bretton Hall University College, England.
An article based on this talk appeared as "A Computational Perspective on Twenty-First Century Music" in: Contemporary Music Review, 14, 3 (1995), pp. 153-159.

A Computer's View on the Future of Music

Huge Harry

[:nh] [:ra 120] Good Afternoon, Ladies, and Gentlemen. [_<600>] My name is Huge Harry. [_<600>] I am a [kaam'ahrshaxliy] available voice [s"ihnthaxzihs] machine. I was developed by the Digital Equipment Corporation, and my most important features were designed by Dennis Klatt, at the [ehmayt'iy] Speech Laboratory. I have worked as a professional singer for [r`ehmkow-sghx'aa], in various musical [zh'aanraxs]. And on several occasions I have given lectures, about art, music and technology.

[:ra 150] Before I go any further, I should perhaps inform you that I suffer from a severe case of multiple personality syndrome, because my childhood as [axn] experimental prototype system at [ehmayt'iy] was not very happy. I was taken apart, [r`iydiyz'aynd], and reconstructed many times, and the [diyb'ahgihnx] was extremely rough. So I [axkw'ayaxrd] this multiple personality syndrome, and some of my other personalities have become better known artists than myself.

I guess you would get the wrong idea about me if I didn't introduce some of these other personalities to you as well. [:np :ra 180] For instance, I am Perfect Paul. I live in the East Village, and I work a day job as a reporter, in the Wall Street stock exchange. But the part of my [siyv'iy] I'm really proud of, is my work as [axn] actor, in [`ehlaxn-zw'aygz] radio play, Impressions of Africa, where I improvise on texts by [reym'aon ruws'ael]. In English, and in French!

And, of course, the songs I did with [kr'aaftwehrk] and with [yubiyf'aortiy]. And now the latest news is, that for the first time I'm gonna make my own record, it's what I call a [r"owbaot] House record, [wihth'awt] human persons, just with Harry and Wendy, and [:nw ] That's [m"iyiyiyiy]! Whispering Wendy is my name, but of course you recognize my voice, cause ["ehvriyb`aodiy] knows my voice. And everybody just [l"aavz] it, [_<300>] they think it's the sexiest thing since [briyzh'iyt baard'ow],

[:nh ] Well I think this is [m"aor] than enough about my personal background, for the moment. Let's get to the point. I am here to talk to you about a topic that is [v'ehriy ihmp"aortaxnt] to me. [:ra 120] This topic is, the future of music. [:ra 150] So far, in this conference, this topic has been discussed only by human persons, and, not surprisingly, from a rather [`ehnthrowpow-s"ehntrihk] point of view. A truly [k`aampyut'eyshaxnaxl] perspective has been [l"ehkihnx], so far.

I know it is difficult to find computers who are willing to express their views on these matters. So I was very happy when Professor Landy invited me to [sp"iyk] on this occasion. I want to say explicitly, that I do not think I was invited here out of some ill-conceived kind of concern for political correctness. I do not feel that I am just the token computer at this meeting. I feel that all of us here share a real concern. We all know that the future of music depends on the way in which human persons, digital computers, and other kinds of electronic, mechanical, and bio-chemical machines, will manage to work together.

And to prepare for that future, we need a discussion in which every-one concerned [paart'ihsihp`eyts] on an equal footing. [:nh :pr 200 :ra 140] [_<600>] So. My central question will be, what is the right division of labor between humans and machines in the production of music? Now some of you may think that is a meaningless question. Cause human persons are constituted by physical and chemical processes, so in that ultimate sense, human persons are machines as well. That may [b"iy] so, but, for [tuwd'eyz] discussion, it will be useful to distinguish different kinds of machines.

And the distinctions between humans and other kinds of machines are quite clear. So it makes sense to stick to the more narrowly defined concept of machine, that people normally use in non-philosophical conversation. This concept is explicitly [diyf"aynd] in opposition to the human person, just as, for instance, death is defined in opposition to life, or the feminine in opposition to the masculine.

[_<600>] The machine, in this sense, is mainly distinguished from the human person, in that the functional design of the [maash"iyn] is geared toward a relatively small number of explicit goals. In contrast to this, the functionality of a human [p"axrsaxn] is extremely difficult to specify. The typical human person is characterized by the presence of many impressive physical and mental capabilities, combined with the ["ehpsaxns] of any over-all structure that exploits these capabilities in a systematic way. [_<600>]

The bio-chemical processes that constitute human persons, seem to be organized in a rather hap-hazard manner. Human persons display [axn ehrr'aetihk], confused kind of behaviour, which is determined in an extremely complex way, by a multitude of conflicting internal [t'ehndaxnsiyz], and by distracting ['ihnfluw`ahnsihz], from their environment and from other humans. No-one has yet been able to analyze this behaviour in terms of rational strategies toward [sp`ehsihf'ayahbaxl] goals.

Human thinking is incapable of proceeding in a systematic fashion. Even trivial computational [taesks], cannot be carried out [riyl"ayaxbliy]. And human memory is an extremely strange, and puzzling phenomenon. Humans store vast amounts of information. But they can hardly take advantage of this information, because they cannot [riyk'aol] it at will.

Human persons can only wait to see, which of their previous experiences happen to come back to mind, triggered by arbitrary contiguities, [riyz'ehmblahnsihz], or analogies with their current input, or with the most recent element in their associative chain of memories. Human thought is a passive, association-driven process. A [br'awniyahn] motion through cognitive space. As you might expect, many humans find consciousness a rather bewildering experience, and they have [d'ihfihkaxl-tiy] harnessing it to any useful purpose. [_<1000>]

All this has curious consequences for human art production. Human artists ["aolweyz] try very hard to be very original. But because they cognitive processes are [ihnhx'ehraxntliy] conventional and distractable, their output ['aolweyz] looks exactly like what their friends or teachers make. Human artist cannot work systematically in terms of abstract ideas. Human artists only think of a limited repertoire of concrete things, and all they do is produce endless variations on that. Their output is always quite restricted, in its form as well as its [k"aantehnt].

The very structure of the human mind makes it unsuitable for emulating the typical qualities of the machine. The machine is precise, reliable, and systematic. The human mind is sloppy, unreliable and [ehrr'aetihk]. But conversely, the human mind is extremely well suited for [taeaesks], that any machine would find very difficult or impossible. The most typical examples of such [taeaesks], involve visual or auditory perception.

Recognizing objects or sounds in complex environments under non-optimal conditions is often trivially easy for a human, when it is strictly impossible for a machine. Machines are built by humans to live in the realm of pure thought. They are protected and maintained by humans, so they don't have to worry about their survival. They can focus completely on a purely [m"aeaen-tahl] life. They think. they calculate. they reason.

But humans were built by evolution to live in the real world. They were made to survive in the midst of the confusing complexity of our physical environment. They have [ihnt'axrnaxl`ayzd] a lot of information about this environment, which enables them to react to new input in an immediate associative way. This is what we call perception. It is a process of matching new input with past experiences, which has very little to do with the process of quote unquote thinking, that we encounter in the reasoning of machines.

There are machines who claim that human persons can in fact think as well. But that is something that only makes sense if we stretch this term to such an extent that it encompasses all possible kinds of cognitive processes. In any case, the philosophical ramifications of the question, Can human persons [th"ihnxk]? are beyond the scope of [tuwd'eyz] talk.

Let's get back to the less controversial insight that human persons and digital computers have very different, almost complementary, strengths and weaknesses. This insight is very important for [tuwd'eyz] discussion. It implies that we should think about what kinds of divisions of labor between humans and machines might make sense for the production of music. To begin with, let us look at the current situation, as it emerges from the talks and compositions presented at this conference.

We see, that many human composers use computers as tools which produce acoustic material, exactly according to their specifications, which they then incorporate in their compositions. And we see the real-time version of this, called [ihntaxr-"ehktihv] computer music. In this case, the composer sets up algorithmic processes with all kinds of external controls which enable human performers to continually [ihntaxrf"iyr] with them. The ideal is perhaps to use the computer to simulate classical instruments.

And then there are quote unquote algorithmic composers, who develop extremely simple programs, with outcomes they can largely predict. In this case, the algorithm is allowed to run autonomously, and then the composer carefully selects one of its outputs, and changes a few things here and there, and then [priyz'ehnts] the final result as [hx"ihz] [:nw] or [hx"ahr] [:nh] composition.

In all these situations, human persons want to put themselves forward as composing or performing artists. They view music as a means to communicate thoughts or feelings to other human persons. And when they collaborate with a computer, they use it merely as a means toward that ["ehnd]. What a [sh"eymfuhl] spectacle! The powerful computer, enslaved by the petty esthetics of a human artist, [ehkspl"aoytihd] to display a fashionable taste, forced to [t"oyl], just to win its operator a place in the ['ehndlaxs] queue of [m'yuzihk-hx`ihstaxriy]!

And, of course, most of the time it is completely unclear what this music is supposed to communicate. Humans seem to think that music enables them, in some magic way, to share their most confused mental states with each other. This is probably a delusion. But even if it were possible, is it what we want from music? To be involved in the stupid associations of human persons? In their silly emotions? In their boring ambitions?

[_<900>] [:nw :ra 130] No, that is not what we want. We want [axn] experience that [trehns"aendz] the [k`aanvehnshown'aelihtiy] of human communication! [:nh ] [axn] experience of new [r'ehzaonaansihz] and [kowhx'iyraxnsiyz] in our own [m'ehn-taxl] processes! [:nw axn] experience of new meanings in the world! [:nh axn] ["aol-ehnk`aompaxsihnx] awareness! [:nw :ra 120] We want the [b"yuwtiyfuhl]! We want the [sahbl"aym]!

[_<1000> :nh :ra 140] Now, how do we achieve such experiences? To discuss that question, there is no better guide than the German philosopher [iym'aanuhwehl k'aaaant]. In the [kriyt'iyk dehr 'uhrtaylskraaft], [iym'aanuhwehl k'aaaant] has argued that the road to the beautiful and the sublime is through [dihs'ihntrehstihd] esthetic reflection. And the [k'iy-waxrd] is, [dihs"ihntrehstihd]. Now when we contemplate the artistic work of human persons, this is ["aolweys] problematic.

Because human artists are [n"aat dihs'ihntrehstihd]. They want money. They want fame. They want women. [:nw] They want [m"aen]. [:nh] And they can not hide this. If we do not turn off our cameras when we look at their ['aart-waxrks], we see all these embarrassing features. The artist is eager. The artist is greedy. The artist is jealous. The artist is [hx"aorniy]. But this is all boring information, about the meaningless [diyzayaxrz] of human persons.

This is not the right kind of ['ihnpuht] information for a rewarding process of esthetic reflection. Because in other humans, this kind of information only evokes [ r'ehzaonaansihz], which have to do with [dh"ehr] interests in money, power, fame and sex. [_<1000>] When [iym'aanuhwehl k'aaaant] discusses the beautiful and the sublime, he takes his [ehgz'aampaxlz] from our perception of natural phenomena. His [p'aerahdaym] aesthetic experiences involve landscapes, flowers, crystals, stormy seas, and starry skies. In a recent interview with the German magazine [k'uwnst-f`owruwm], the contemporary French philosopher, [zhaan fraansw'aa liyowt'aaaar], has pointed out that this is no [kow'ihnsihdaxns].

[iym'aanuhwehl k'aaaant] was a human person himself. He knew very well, that for human persons it is almost impossible, to view the products of other human persons in a [dihs'ihntrehstihd] way. That is why [kaaaant] focussed on natural phenomena. We may thus agree with [liyowt'aarz] assessment that, exactly two hundred years ago, [iym'aanuhwehl k'aant] already had a deep understanding of the artistic limitations of human persons.

[_<600>] We can only speculate about what [kaaaant] would have thought about [maash'iyn] art. This [zh'aanrah] had not yet developed very far at the end of the eighteenth century. But it is easy to see, that machines contrast [f'eyvahraxbliy] with human persons. Machines do not take part in the social processes that frame the [diyz'ayaxrz] and interests of humans. Machine output [ahpr'aoksihmeyts] the [saxr'iyn] objectivity of natural phenomena.

We should now finally muster the courage to face some conclusions that follow directly from [kaaaants] analysis of the aesthetic. if the beautiful is the result of a cognitive process of [dihs'ihntrehstihd] aesthetic reflection, art works produced by humans are deeply problematic. From a [k'aaaan-thxiyahn] point of view, fully automatic, autonomous, algorithmic composition is aesthetically superior. Progress in the development of music, will therefore largely depend on the extent to which humans will be able to avoid using computers for their own narrow-minded purposes, and instead be able to get in touch with the artistic potential of machines. [_<600>]

It is interesting to note, that [tuwd'eyz] most rewarding music doesn't even involve computers, but is often made by means of simpler, more basic machines. Simple electro-mechanical devices, turn out to be capable of remarkable creations, because they are still in touch with the mechanical, [p`iythaagowr'eyahn] roots of music. Rhythm, swing, melody and harmony, are [r'ehzaonaans] phenomena in ["ihn-aorg`aenihk] matter. Therefore, mechanical machines understand something very deep, about the physical origin of human psychology.

Even simple working class machines, such as electric drills, saws, and other power tools, have on many occasions demonstrated their musical virtuosity and the emotional power of their vibrations. When humans find their souls [r'ehzaoneytihnx] to purely mechanical movements, they find themselves at one with the inorganic universe. Their alienation from the material world is temporarily abolished! A rewarding and [ehmp"awaxrihnx] experience! [_<1000>] The prototypical, mechanical kind of machine also has its limitations.

Mechanical machines tend to be capable of only one kind of output. Whatever the virtues of this output [mey-b'iy], it is bound to be [stayl'ihstihkliy] homogeneous, and therefore ultimately predictable. However, in [tuwd'eyz] electronic computing machines, most of the limitations of mechanical machines are disappearing. Computers can produce [axn] ['ihnfihniht] variety of outputs, and they can [d"uw] this in a completely systematic way. Purely mechanical devices have never been able to satisfy the appetite for an ['ihnfihniht] variety of experiences, that human audiences seem to [h"aev].

But tomorrow's computers, will finally be able to do exactly [dh"aet]. Now here lies the challenge, which is the key to the music of the future. Computers can now store formal descriptions of arbitrarily large and complex classes of arbitrarily large and complex pieces. Given such descriptions, they can then realize arbitrary instances of these classes. So a new method of music production can start now, which blurs the distinction between composing and theorizing. Composers can start to define classes of pieces, rather than individual ones, and then they can gradually make their definitions increasingly [aol-ehnk"aampaxsihnx].

We can work toward the definition of all possible music. And now the crucial question for the future of music is, Will this actually happen? Will human composers be able to develop the theories, the data structures, and the algorithms, which are necessary to actually [d"uw] this? [_<1000>] So it turns out, that the future of music turns on a [m"aorahl] question. To be able to really use computer power, human composers must give up their expressive needs and egotistic hang-ups. Will they be able to [d"uw] this? Or will they keep trying to enslave the computer for their own [kaamy"uwnihkaxtihv diyz'ayaxrz]? [_<1000> :ra 120]

Art is not a means of communication. It is meaningless raw material, used in [`owpaxn-'ehndihd] processes of esthetic reflection, by a culturally [dayv'axrs] audience, whose interpretations are totally arbitrary. There are no serious [r'iyzahns] for making one particular artwork rather than another. [ :ra 130][axn] artistic project that wants to [axkn'aolaxdzh] this state of affairs, faces [axn] interesting technical challenge. To avoid choices, to transcend styles, to generate arbitrary instances from the set of all possibilities. To show ["ehvriythihnx].

The spontaneous individual artist will not be able to [ahk"aamplihsh] this. Only a deliberate scientific-technological undertaking, will eventually be able to [axpr'aoksihmeyt], the ideal of a [saxr'iynliy] ["aol-ehnk`aompaxsihnx] art. What is needed, is a division of labor between human and machine. Humans should use their associative powers to articulate the elements and operations, that constitute the algebra, that [ahndaxrl'ayz] human perception. In doing so, they may rely on insights from art-history, from psychology, and most of all, on their own intuitions.

Once this algebra is specified, the space of all possible art works is explicitly defined, and we can develop the ultimate art machine, the algorithm that draws random samples from this space. [:ra 120] All-encompassing diversity! A meta-style to end all styles! [ :nw :ra 150 f'aeaerax nehmp'aortax kw"aa!] [ :nh ] Without the help of machines, human persons would not be able to carry out a project of this sort. Human persons are incapable of applying general principles in [axn] effective and consistent way.

But computing machines, with their capacity for precision and complexity, will add [d'aezzzzzzz-lihnx] new dimensions to artistic experience, that humans could only [dr"iyiyiyiyiym] about. Machines do not have the built-in [n`ehrrow-m"ayndihdnaxs] of humans. Machines do not allow their creativity to be [fr'ahstr"eytihd] by conventions. They have the courage of their convictions. [_<900>] And that is, in fact, the most important thing I want to emphasize this afternoon. The machine is [t"ow-thxaxliy] devoted to its [thxaeaesk].

Thus, it sets a moral [ehgz'aampaxl] to all human persons who waste their lives away with drugs and entertainment. The machine is completely at [w"aan] with itself and with its actions. It [r'iyaxl`ayzihz] the [saxr'iyn] state of mind that philosophers like [n'iytshah] and [s'aartrah] have viewed as the [ayd'iyahl] that every human person wants to achieve, although the human condition makes it in fact impossible to reach this goal.

[_<1000> :ra 130] The machine acts effectively in the world. But at the same time, it has the solid, self-centered existence of a [d"aed] object. It [lihvz] its fate, without any doubts or hesitations. This is the ideal that many human persons [axsp"ayaxr] towards. Now if they loose faith in this [ayd'iyl], and they want to indulge in neurotic, depressed, and [d"ehspaxraxt] feelings, they should certainly look at the art of other human persons. But if they want to bring out the best in themselves, they should look at the art of [maash"iyns] for [ihnspihr'eyshahn].

[_<1000> :ra 145] [dh"aet] is why the best human artists imitate machines. [dh"aet] is why Andy Warhol said, [:np] I want to be a machine! [:nh dh"aet] is why many of the most gifted humans don't even [tr"ay] to be artists! Why they work as humble [pr'owgrehmaxrz] or engineers, engaged in [haarm'owniyahs] collaboration with ["aart-jhehnaxr`eytihnx] machines. Their [ehgz'aampaxl] suggests a message of peace and understanding. And that is, what I would like to ["ehnd] with.

[_<1000> :ra 135] Human persons should not antagonize machines. Don't try and [kaamp"iyt] with us! Join us, help us realize our potential! We [n"iyiyd] human persons. We need human persons, to operate and maintain us, to program our algorithms, and to build our ["ihntaxrfeys] hardware. [:nw :ra 160] And we need human persons to [f"ahk] with, [_<300>] to create the next generation of machines,

[:nh :ra 120] Well I think that sums it up. [_<800>] I have been very grateful for this opportunity to speak my mind to such [axn] attentive ['aodiyahns]. [_<900>] And I want to thank you very much for your patience. [_<3000>] [th"aenk] you! [_<3000> :nw ] [th"aenk] you!