THOSE MARVELOUS
MUSIC MACHINES

ALGORITHMIC COMPOSITION SOFT-
ware is a little bit like network television shows. (Or maybe the atomic bomb.) Once the concept has been born and the machinery to deliver it is in place, the stuff starts to proliferate. Whether anybody has any real use for it becomes immaterial. Jim Johnson's original Algorithmic Composer begat several other programs from Dr. T's, Intelligent Music upped the ante with their Jam Factory and M, and now we're seeing a host of imitators, from Sound Globs (Twelve Tone Systems) to OvalTune (Intelligent Music again).
One thing these programs have in common is that the computer becomes an active participant in the music-making process. The user gives high-level instructions and allows the software to fill in the details. (The instructions may include randomization, or they may not.) Another thing is that none of them operates within the confines of the multitrack tape deck metaphor. Each program restructures, in its own ways, the manner in which the musician will conceive of and manipulate the music.
Now here's the kicker. There's a reason why most sequencers are designed around the metaphor of the multitrack tape deck: It's a metaphor that meshes closely with how we think about and record music. So when a music machine comes along that departs radically from this metaphor, we have to ask what it provides in the way of alternatives. How does it allow (or force) us to think about music?
We're borrowing this rap from Bob Moog, who made a similar point about synthesizer programming. What we want in a synth, he said, are controls that give us rapid, precise control over expressively significant aspects of sound. In the same way, when a computer becomes a music generator rather than a blank slate, we need to ask whether the dimensions of musical control made available by the software allow us to reach in and easily make changes in the control structures that will have expressive results. "Expressive" doesn't have to mean "familiar ~' -we're not talking about automated three-minute hit singles here. But the changes should be intuitively meaningful in terms of the ways that we listen to music, and they should be arrived at through methods of interaction that computer-literate musicians, at least, will feel comfortable with.
That's a tall order. The "conductor grid" in M is a rough stab at a new metaphor for live interaction, but it's probably too amorphous, and certainly not finely shaded enough, to really do the job. We haven't looked at MidiDraw yet, but those gross squiggles in the ad don't look at all the way we imagine that we'd like our music to sound. Music Mouse and Mousterpiece offer a comfortable, but limited, interface through which gestures can be fed into the computer. The gestures are used only to generate notes, however.
We're moving, slowly, in the direction of a music system that would combine two elements: a flexible, expressive performance interface (probably a hand-held alternate controller) and a computer program that will interpret the data from the controller. The program won't simply remap and play back notes, though it might do that. It will start pre-programmed events in response to cues from the performer, and will alter, in expressively meaningful ways, events that are currently being played back.
One more little detail: This controller computer system will have to exist in a stable form for long enough that musicians can develop performance skills and resource files for it. Sure is a nice dream. Maybe someday.


Keyboard Magazine