Talking Music with a Machine

  • PDF: https://www.jstor.org/stable/40375093

The interactive computer-controlled music box we are currently developing at York represents yet another attempt to harness the formidable power of the modern digital computer for the improvement of education. After years of such experiments, many of us are understandably skeptical about extravagant claims that computers can somehow make the teaching process more responsive to individual needs and abilities. Hence the practical focus of our research. By putting in two years of intensive testing, first within University walls—where we can benefit from a wide range of expert advice and technical help—and later through visits to neighbouring schools, we want to find out whether, this time around, such hopes will prove more realistic.

Ours is, in fact, one of the first computer terminals to be specially adapted for elementary work with music. Fortunately, it is quite a bit simpler than many earlier computer-driven music machines, in its hardware at least. This means that most students can, without special training or background, learn to write and edit their own music-generating programs after only a short acquaintance, rather than depend on special teaching programs laboriously prepared in advance. Such a system, we have found, lends itself to a wide variety of simple experiments in composing and decomposing tunes, rhythms, and other musical patterns, even including chords and counterpoint. We are now trying to bring it past the gadget stage, to see whether something serious and useful really can be learned from such an "instrument," something that normal methods cannot or do not teach as well.

Of course, the very idea of using a machine to teach music—let alone one with an "electronic brain"—seems threatening to some lovers of music and of children. For one thing, it challenges the reigning belief that ensemble performing and listening are the only practicable vehicles for general music study, at least in the school setting. On the other hand, computers are exciting to kids. And many of today's leading music theorists, cognitive psychologists, computer scientists, linguists, and teachers of design are already attempting to use computer-generated models to gain new understanding of symbolic processes and mental development in children, musicians, and other humans.

So, to help clarify my own thinking on these issues, and to stimulate interest in problems of teaching and learning among my students and colleagues, I set out three years ago to get my hands on some sort of interactive computer learning system and put it to work on musical problems at York. After much digging and several false starts, I was fortunate enough to receive a grant from the Planning and Research Branch of the Ontario Ministry of Education, which was looking for a preliminary firsthand assessment of what such systems might eventually have to offer the schools in this Province. The York Interactive Music Project is now in its second year, and so far, one thing at least is clear: this machine is going to teach us plenty, whatever else it may or may not do for children.

As a starting point, we have taken an existing high-level computer language called LOGO, designed for educational use. It is based on familiar English words, and makes it easy to define and add new words to its vocabulary. In more than a year of painstaking trial-and-error, we have used this basic code to construct a fat dictionary-full of special music-oriented programs, aimed at the novice user or would-be composer—in effect, a whole new LOGO dialect. Thus a tool originally designed for computation and mathematical problem-solving has been converted into one that can pose musical problems, while encouraging each student to construct his own "solutions."

Jane, for example, will conduct her musical dialogue with our computer via a typewriter-like console, a standard model with a few small gadgets attached. After a brief exchange of typed introductions, she can begin to write notes—an integer to express each discrete pitch, with another integer for each duration. (Normally, pitches are numbered consecutively in chromatic steps over a range of five equal-tempered octaves. But degree-numbers can be used instead, the reference scale being chosen or created by Jane herself. Or, she can make her numbers represent intervals rather than actual pitches—whichever scheme seems handiest.) As soon as she finishes typing each sentence of numeric code, the numbers become sound—hardly more than an electronically-beeped outline of the pitches and rhythms she specified, but still quite recognizable. Easily replayed on demand, the heard tune makes Jane realize just how close her typed symbols came to representing the result she imagined: for the computer prints out her numbers (or a simple graphic equivalent) and keeps them in view while it plays the corresponding notes on the music attachment box. After further trial-and-error, she smiles as a good result finally sounds from the box, and immediately puts a label on it for future reuse, by typing CALLIT BEEP1 (any handy name will do). From now on, BEEP1 is all she needs to evoke that particular tune.

After a while she is no longer just typing sentences of numbers, but working mostly with the names she has given to significant musical components—motives, pitch sets, interval patterns, rhythms, whole phrases, anything worth identifying and repeating. Eventually, a whole section of Jane's new piece is ready for naming. Growing naturally from the musical units she built up and christened—some of which she may have saved from previous encounters—it was made by linking, super-imposing (chords or counterpoint can be played in up to four voices), or otherwise changing and regrouping them. To work these changes, Jane was given a handful of ready made operator-words at the outset. But she may decide that some new way of processing or changing tunes is needed also, to speed her work. If so, she can add the new procedure to the system herself, writing it in the form of a generalized recipe, which goes into action whenever the codeword for her operation precedes the name of a musical unit in her typed instructions to the computer. (Thus, SCRAMBLEPITCH BEEP1 might become her own formula for randomly reordering the pitches in a tune—BEEP1 in this case—while keeping its rhythm intact). In time, of course, her working vocabulary of operations and musical building-blocks could grow unwieldy, unless she can find some logical system for keeping tabs on everything—a cognitive lesson that any maker, any designer, anyone attempting to work within a complex structured world, can profitably take to heart.

Shaping words and numbers into a small symbol-world of one's own, then using these symbols in various combinations and hierarchies to create new melodies and rhythms, hearing and judging the results for oneself, is an engrossing experience for almost all ages. Indeed, it reminds some people of the process by which children learn to use ordinary language—perhaps because both depend so much on the desires and ideas each individual brings to the task.

At the same time, working with our system seems to provide a natural incentive to logical thinking—particularly if we remember that, as Christopher Alexander has said, "Logic, in the widest sense . . . is concerned with the form of abstract structures, and is involved the moment we make pictures of reality and then seek to manipulate these pictures so that we may look further into the reality itself."1 To be sure, if one's main interest as a teacher were in programming the child to respond to our questions when we want an answer—in approved verbal formulas that may have little to do with the child's own mental image of the shapes or orders he has actually heard—then the educational value of playing musical games with a machine like ours would no doubt be harder to accept. Whether, and in what settings, such a system can prove its value to children remains to be explored in the next phase of our research.

Since last Fall, York students have been using the computer music system for a variety of projects, including an experimental Humanities course I give on Order and Structure in the Arts. In its short life at York, the Interactive Music Project has already stimulated all of us who have come in contact with it—and particularly the students themselves—to become much more aware of the hows and whys of learning. Teaching the machine and each other, trying to rethink musical operations in procedural terms, has made us more receptive to issues of method and strategy wherever they appear. And as we meet more often with teachers in the Toronto area to demonstrate our system, certain disturbing questions have come to the fore with particular persistence, making us look more and more critically at what we are trying to accomplish. Here are some of them:

1) Do we insist too much on musical notation as the medium through which all awareness of form and design in music must be gained? Does not conventional notation often impede our grasp of structures and relationships, even when we have become habituated to using it as a guide for performance?

2) Can we really increase musical awareness by teaching a vocabulary of verbal labels, however carefully keyed to prerecorded excerpts or analytic charts? Or should we look for new ways to help each listener articulate his own sense of what he hears, even before asking him to represent it in words or notes?

3) Are exercises in motionless, contemplative listening the only, or even the best way to arouse and engage a young person's musical imagination? Can the kind of organized performance activity we call "practice" or "rehearsal" be expected to do the job, when so much attention must necessarily be given to techniques of sound production and deciphering written instructions?

4) How do different kinds of musical learning fit best with the stages of a child's development and schooling? There seems to be general agreement that learning to sing belongs in elementary school, learning to play in the middle years, learning to talk about music in college. But what about the other symbolic, analytical, constructive skills music has to teach our children? Are these being given too short shrift? Can the experience of building and manipulating simple musical structures be made more accessible to ordinary kids in ordinary schools on an everyday basis?

5) Can a computer program tell us anything we do not already know, either about the structure of a musical phrase, or about how it was composed? Is not computer language too abstract, too schematic, too literally precise to handle any really significant or interesting relationships? Is the computer not in fact dangerously anti-musical, by encouraging students to think of music in terms of abstract operations and numbers, rather than personal expression and sensuous beauty of sound?

I sometimes wonder, in all candor, whether the business of teaching music, still so much a matter of instinct and tradition even in this era of computers and moonprobes, can safely withstand the sort of probing question this particular computer seems to elicit from those who have seen it in action. Yet there are signs that suggest our project may not be completely misguided or premature.

For one thing, a machine that represents combinations and transformations in recipe form, and lets us test these recipes in audible sound, should offer a natural base for experiments in the psychology of musical perception—a field that has long been dominated by nothing more interesting than the measurement of statistical responses to the stimulus of isolated tones. One psychological issue of special importance to all artists is the relative weight of conscious, explicit principles of design and relationship to unconscious or implicit ones, in the work of making forms. Seldom have we had a tool that is so well adapted to shed light on this problem. Not only is a system like ours remarkably conducive to the systematic trial-and-rejection of alternate solutions, a process so important in all the arts of design. It can also enable us to follow and record many of the steps traversed and choices made in a complex sequence of musical invention, for later analysis by students of creative behaviour.

Such a procedurally-oriented instrument might also be of considerable help to theorists and musicologists, as well as to ordinary students with little skill in physically executing music (some of whom, alas, inevitably become theorists or musicologists!). If only by inviting us to translate hitherto vague notions into language precise enough for a machine to grasp, it could help illuminate the relationship between analytic ideas and constructive techniques in new and powerful ways. If such tools ever do become widely accessible, they will spur us to bring the teaching of music, at every level from kindergarten through university, into a closer and more fruitful relation with other branches of abstract formal thought. This is, of course, a direction in which some of the best of living composers have already begun to move, following a tradition as old as Pythagoras. Can we teachers afford not to join them?

By the time systems such as ours are ready for full-scale trials in the schools, the rapid evolution of miniature computer hardware will likely have created a whole new range of technical possibilities. Costs, too, may then be reduced substantially; more important, the physical size and manageability of computer learning tools may for the first time approach that of, say, a closed-circuit TV system or a concert grand piano—possibly smaller and handier still—and could come within reach of the ordinary local school board, community arts center, or smaller college. But when that day comes, Ontario schools and colleges will need people already on the scene who know how to make intelligent use of what technology has dropped in our laps. This is one reason why I set out to recruit a research team of working schoolteachers, college students, and interested faculty experts for this project. Together, we are exploring the educational potential of York's computer music box while we can still hope to guide its future development.


1Notes on the Synthesis of Form (Cambridge, Mass.: Harvard Univ. Press, 1964), p. 8.

Read 1124 times

Last modified on Monday, 12/11/2018

Go to top