An illustration in Ray Kurzweil's 1999 robots-are-coming manifesto The Age of Spiritual Machines depicts a desperate-looking fellow, complete with sweating brow, laboring over a writing pad in a bare, bulb-lit cubicle. Strewn about the room, discarded sheets of paper bear such claims as "Only humans can play master chess" and "Only humans can recognize faces," each heavily crossed out. Among the casualties: "Only humans can improvise jazz." Yet another endeavor once thought to be our own exclusive cognitive province has, it would appear, been colonized by faster, smarter, ever more complex computing systems. It's the kind of dramatic assertion AI boosters are fond of, and one that couldn't be better calculated to shake one's faith in our unreplicatable specialness. Even more than most creative endeavors, jazz is surrounded by a rhetoric of "intuition" and "inspiration," especially with regard to the central role of improvisation. This is partly historical: Many of the music's heroic figures have been longer on lived experience than formal musical training. But it's also because the very idea of creating moving (or even coherent) music "on the spot" or "in the moment" can seem like some mysterious conjuring trick, both to non-musicians and those schooled in the European classical tradition, with its strict separation between composition and performance. Some jazz greats have played up the popular image of their activity: "If you haven't lived it," Charlie Parker famously said, "it won't come out of your horn." On the other hand, avant-garde saxophonist (and MacArthur fellow) Anthony Braxton has written witheringly of "the myth of the sweating brow" (the idea that a soloist is only really blowing if he appears to be undergoing inner tumult) and "the myth of the good night" (the idea that a quality of a jazz performance depends more on fate and circumstance than on skill or creativity). For Braxton, these notions -- especially when applied to black musicians by white critics -- devalue both the brainpower and long experience needed to create good improvised music. Whatever your view of jazz, one thing is clear: Computers don't have "good nights," and their "lived experience" is, shall we say, limited. That's why the idea that one might extemporize with human speed and fluidity is such a fascinating one. Such a program, assuming it passed musical muster, might overcome the long-standing downside to recorded jazz, which "freezes" a single performance into an infinitely repeatable form. Our hypothetical Deep Blue Monk might, on the other hand, produce a fresh rendition of "Round Midnight" with every click of the mouse -- and never show up late for gigs. But don't throw your copy of Brilliant Corners out just yet. Some of the existing projects in the field fail to live up to Kurzweil's billing, while the most interesting don't even try.
What might make a program count as a full-fledged improviser? First, by definition, its output would have to take place in real time. Though only the furthest-out free-improv purists claim to have no preconceptions about what they're going to play ahead of time, a computer (or musician) that thoroughly composes what it's going to perform, and then performs what it has composed, is no improviser. Second, the program would have to allow for interaction between computer-generated output and other musical elements. There's a tradition of solo improvisation within jazz, most obviously among pianists, but the feedback loop among players responding to one another's moment-to-moment decisions supplies much of the music's interest. The project most-often cited (by Kurzweil and others) as filling the bill is called, naturally enough, Improviser. The brainchild of British saxophonist and programmer Paul Hodgson, Improviser's most public gig to date was on a 1998 BBC segment, trading Parkeresque licks with noted jazzman Courtney Pine. These were hardly Turing Test conditions, and Pine's assessment of this digital duet was vague at best: "You can't replace the human experience, the human spirit." Unfortunately, both the BBC'S account of the session and a brief description of Improviser in a 1996 Discover article on computer creativity, by AI-Life specialist Margaret Boden, are short on details: Does the program's output sound like a saxophone, or is it a MIDI-voiced approximation? Does it replicate Parker's attack and phrasing, or only the timing and selection of notes? Most importantly, does it interact with other musical input -- and if so, how? David Cope, the composer and programmer behind Experiments in Musical Intelligence (EMI) has heard the fruits of Improviser's labor. "Paul and I have sat down a few times and played each other tapes," he told me, "but I've lost track of him. At first swack, his program is really quite good at producing believable licks in a Parker style. It sounds like an improvised solo, but there's no ending or beginning, and even the most improvised music has points of arrival and direction. He hasn't yet had a chance to pay attention to the way a piece works. I liked what he did, but he agreed that the program needs quite a bit of work." Cope's assessment carries weight: His own work might be thought of as the classical-music analogue to Hodgson's jazz-based efforts. Since its beginnings in the early 1980s, EMI, which Douglas Hofstadter has called "the most thought-provoking project in artificial intelligence that I have ever come across," has produced thousands of "recompositions" in the styles of everyone from Bach to Rachmaninoff to Cope himself. In three books about his work, each accompanied by CD-ROMs containing simpler versions of the code he works with, as well as audio samples of EMI's output, Cope offers both technical explanations of EMI's architecture and coding (in LISP) and speculations on its philosophical implications. The latest, the Algorithmic Composer (2000), focuses on Alice, a program that acts as a composer's assistant by applying EMI's principles to work in progress. How does EMI work? First, Cope codes suitably similar music by a single composer as MIDI information. (The rigid structures of Chopin mazurkas or Bach chorales give the most impressive results.) The resulting database is chopped up into beat-long units, which are stored with both their harmonic character and their place in a composition's structure. "You could look at the information each piece contains," Cope says, "This piece has three simultaneous lines, here's where the lines were headed, and it's the second chord of a five phrase section of a four section piece." Additionally, a pattern-recognition subprogram uses music-theory criteria to identify characteristic "signatures" common to the body of work at hand. (It's this feature that lends some of EMI's output its spooky feel of being the handiwork of a single composer with a recognizable style.) EMI then proceeds to recompose, by first selecting something it recognizes as a beginning, and then choosing subsequent units that make a snug structural and harmonic fit with what has come before. "The best choice is often the one the composer picked, and that will be nullified," Cope says. "It's also got to be sensitive to the fact that this is the sort of thing that the composer would do at the same time in same sort of place. It can't do certain things in the middle of a cadence, for example." At the level of overall form, the program borrows a page from classic AI work in natural language processing (NLP). Just as a sophisticated NLP program can derive the sentence "John was loved by Mary" from "Mary loved John," EMI applies what is called a "recursive transition network" to a piece's underlying grammar, eventually transforming it into a new one. The musically informed analysis upon which EMI bases its transformations is considerably subtler, but the program is ultimately rule-governed in a similar way. Cope estimates that current versions of EMI can write about "two hundred bars of music a minute," though coding the source data can take considerably longer. (The output can either be played as a MIDI file or printed out as a score for human performance.) But speed is not the main aim of the project, which is firmly rooted in Cope's compositional background: "Very early on, I just said, 'I'm not going to strive to do this in real time.' Still, one can see how EMI could be applied to the goal of producing convincing improvisations in an extant jazz style. The bulk of swing and bebop consists of variations on a set song structure, most often one in thirty-two-bar AABA form, and the melodic content of a solo by, say, Parker is to some degree constrained by the song's harmonic content. Writing an algorithm to play such solos might be a matter of chopping up a body of performances into scales and patterns, and supplying instructions for how and when to apply them in a given harmonic context. Recent versions of EMI compose a piece's phrasing and dynamics along with its pitches and rhythms; there's no theoretical block to doing the same for jazz. But even with real-time performance (which might be simply a question of computing power), a single solo voice enslaved to a predetermined rhythm section accompaniment would seem to be insufficiently interactive to count as improvisation. (As it stands, EMI actually composes all of a piece's voices at once, making the notion of a "solo" fairly meaningless.) Also, Cope is quick to point out that much of what gives jazz its character would be harder to mimic. "One of the most important things about jazz is its rhythmic fluidity, and one feature of EMI is it rhythmic rigidity. It has to know exactly when the last piece was so it can match the next piece to it." The samples he's heard of Paul Hodgson's Improviser run afoul of the same difficulties: "He formed his database from notated music, and then applied an algorithm that added the phrasing, so it had the rhythmic earmarks of a jazz performance style. I disagreed with that approach. I would want to find a way to chop up non-notated music in such a way that the performance is preserved." Several projects are higher on Cope's agenda, but EMI may tackle jazz yet: "I've collected about 150 like-styled MIDI files from the Internet to use as a database. It's the worst possible generic slop, in a general piano style. But it does have the stylistic and rhythmic subtleties, the notes 3/1000 of a second after the beat or 4/1000 of a second before the beat. And I've written an in-house document that -- I believe -- sets the stage to write the code for a jazz improviser that would produce the same quality of music as the classical program. But I don't know when I'll have the hundred or two hundred hours it would take to write the code."
If EMI's raison d'ętre lies in its emulation of established styles, the question of whether it might do for (or to) Parker what it's done for Bach begs larger ones: Is this all there is to jazz, the music famously characterized as "the sound of surprise"? Is the activity of improvisation nothing but rapid composition within generic limits, complete with signature licks and a certain rhythmic profile we're pleased to mystify as "swing"? Trombonist/composer/programmer George Lewis's interactive system Voyager, which produces unpredictable real-time responses to human improvisers' input, bears directly on these questions. "When you listen to it," says Lewis of the music Voyager helps create, "it's not going to sound like baroque music or bebop. It doesn't sound like something that you can put a label on, other than free improvisation, which is more a methodology than a genre." Lewis's performative aesthetic was formed in the Association for the Advancement of Creative Musicians (AACM), the Chicago 'free music' collective that served as a springboard for the Art Ensemble of Chicago, trumpeter Wadada Leo Smith, and Anthony Braxton, among others. Conceived in the early 1980s, during a sojourn at IRCAM, the French electronic-music research institution founded by Pierre Boulez, the Voyager project stemmed from Lewis's reflections on what he wanted, as a player, from his improvising partners: "An improviser is playing their own music, making decisions on their own. To get a handle on what that process was, I tried to create some device that would do some of the things that I thought were necessary to bring it into a space with other improvisers. I wanted to create something that would make human improvisers feel that somehow they were being understood or listened to -- that, somehow, there was a perception of dialogue." The resulting system has gone through numerous incarnations in the last twenty years (the current version is written in Formula, Ron Kuivila's music-friendly variant of Forth). It receives input from a device called a pitch follower, which translates what a musician plays into information usable by a MIDI system. (Think of it as a kind of analog-to-digital microphone.) It takes a bit of work to turn this data into something bearing the sort of information that Lewis wants the system to take into account in creating real-time responses. "What the pitch follower gives you are onset times -- when the note goes on and off -- how loud the note is, and some minor notions of contour you can get from that," Lewis says. "That's what the machine gives you for free. After that you have to create an analysis program for a set of features that you look at over time. If you know when pitches go on and off and on, they you know about the amount of silence people are using, and the kind of notes that people are playing." Voyager's interactive elements come in three main varieties: "The response will run the gamut from trying to imitate you in some way, to working against you, but you still know it's taking your information. Then there's a third mode -- and these are very crude approximations -- where it doesn't pay any attention, and you have to decide what to do with that. If you decide to be silent and let the machine take it, the machine can take it." This automatic composition element, which includes a significant amount of randomness, lays much of the responsibility for the shaping the encounter at the human improviser's feet. And once a performance is initialized, "there are no buttons you can push to make it do anything -- it's not an instrument." The current version has an output of sixty-four parallel voices, organized into differently-timbred groups that may be responding to input in different ways at any given moments. The system's range of response is complex enough that even Lewis says he doesn't always know exactly what it will do, though he can often predict when a major change in texture is coming: "There's a lot going on, so sometimes it's hard to tell what mode it's in."
Lewis' boldest claim for Voyager involves its capacity for what he terms "emotional transduction." In "Too Many Notes," an article in Leonardo Music Journal (the MIT-based Meatpacker's Annual of the electronic-music community), Lewis suggests that a Voyager performance constitutes a "subject-subject dialogue," in that the computer-generated element is a response to features of the human performer's choices and decisions, as well as the other way around. As a result of this two-way conversation, Lewis writes, "the emotional state of the improviser may be mirrored in the computer partner," even when the latter isn't mimicking human input in musically straightforward ways. In conversation, Lewis doesn't sound nearly as ghost-in-the-machinish. "When I say 'mirrored,' the question is who is the mirror? Obviously, it's us. What I was trying to talk about was the idea that sounds are carriers for other kinds of meaning, which seems undeniable on the face of it. Even two people jingling the same set of keys seem to be discernable. So the idea is that if sound-producing behavior can induce emotional responses, then a behavior that is somehow detected by a computer and reinjected back into the space in a different form might do so as well. I'm not making a claim-that's the business of science. I'm proposing this as an artifact of my experience." Put this way, the idea of "emotional transduction" isn't essentially computer-related at all. We needn't ascribe a will to Voyager -- any more than we normally do with a camera or a tape recorder -- in order to find its representation of human-made input expressive. By now, it should be obvious that neither EMI nor Voyager is intended to supply a straightforward answer to the question "Can computers improvise jazz?" Cope and Lewis share a cautious attitude toward calling their work artificial intelligence; both seem more interested in whether their systems can create worthwhile music than whether they do so in ways that parallel human cognitive processes. Nonetheless, their projects reflect very different ideas about musical creativity, ideas that in turn reflect their creators' respective backgrounds in composition and performance. (Despite this, they have played joint concerts on occasion, and remain collegial -- when I tell Cope that I'm interviewing Lewis, he sends along his greetings.) EMI operates according to Cope's hypothesis that a great deal of what we take to be the ineffable genius of classical masterworks can be largely captured by careful analysis and syntactic manipulation. Voyager, meanwhile, offers an environment that, in some sense, "takes seriously" a performer's intentional choices, consonant with improvised music, people are generally using will creatively, and are interested in what it can do."
So whither Deep Blue Monk? Lewis, as you might suspect, has little interest in building a system that can pass the jazz equivalent of the Turing Test for reasons recalling the comments of Braxton and Bailey above. "Passing a Turing Test is kind of irrelevant to the making of artwork that assumes that there is going to be a single stable standard for what intelligence is, how it's constituted, and how it's measured. But there are people searching for extraterrestrial intelligence, and some of them wouldn't recognize Albert Ayler as music. So if Albert Ayler came floating through their machines, they wouldn't recognize it as the product of intelligence. I can't subscribe to that, just for reasons of my own education and cultural background." Even Cope, who has spent the last twenty years on its classical counterpart, questions the point of such a project: "If you're imagining something like a Dixieland jazz quartet, with the computer as the trombone, that would be an extraordinarily difficult task. A lot of brilliant minds would have to make an enormous investment of time and energy. And I don't know what would be achieved by that, beyond novelty or shock. I think the reaction would be, 'That was interesting -- now let's hears some jazz again.'" Such self-doubt about the viability of one's endeavors is only human -- for the moment. Franklin Bruno writes about music for CMJ Monthly, Time Out New York, and other publications. He plays and records his own songs under his own name and as one-third of Nothing Painted Blue. He lives in Los Angeles.
Other articles by Franklin Bruno
MetaLinks
George Lewis\'s interactive system In his 1980 book, Improvisation, free music guitarist and theorist Derek Bailey complained that a growing emphasis on formalizable aspects of jazz had led to the ossification of the music as a creative force. \"Taking the music made by, say, Jack Teagarden or by Albert Ayler and extracting it from a \'method\' is difficult to imagine,\" Bailey wrote. \"On the other hand, bebop has obviously been the pedagogue\'s delight. It has proved to be one style of improvising which can be easily taught.\" If creating jazz is a matter of stylistic correctness, then Cope and Hodgson\'s syntax-driven approaches might seem to be on the right track. But if it\'s the give-and-take among players responding to one another that makes jazz more than the musical equivalent of doing cube roots in your head, EMI is a non-starter.
among others In addition to Lewis himself, Douglas Ewart, who performs on homemade wooden flutes, and the AACM saxophonist Roscoe Mitchell have performed with Voyager, as have such adventurous English musicians as violinist Jon Rose and saxophonist Evan Parker (who is sort of the anti-Courtney Pine).
sometimes it\'s hard to tell what mode it\'s in The computer-assisted improvisations heard on Voyager (1993 Avant, Japan) and Endless Shout (2000, John Zorn\'s Tzadik label) may not conform the typical
listener\'s idea of jazz, but they\'re not without musical precedent. The actual sound choices range from realistic MIDI percussion to bleeps and whooshes that read as \"computer music,\" but the overall feel is pointillistic and orchestral, not dissimilar to a many-membered AACM ensemble, or one of Butch Morris\' large-scale \"conducted improvisations.\" The strategic push-and-pull at the heart of the program has much in common with the gamesmanship of such European free improvisers as Bailey or Misha
Mengelberg. And in its atonal but melodically contoured character, its responses seem best suited for players whose improvisational decisions are less about pitch selection than timbre and texture -- players, in short, like Voyager\'s creator. This isn\'t entirely accidental: \"I think of this as being my music. It\'s my conception of how I would write if I were interested in hearing a symphony orchestra improvising.\"
Back to Article
|