I’ve funded Jazari without grants, which means I’ve never had to put together a proposal in which I justify my project with references to obscure aesthetic philosophers, giving back to the community, or helping children overcome disabilities (I could see some potential in the last, actually). Being freed from the constraints of the grant process has saved me a lot of time and the world a lot of bad prose, but it’s left me under-prepared to answer challenges like “So, what’s the point?” Here, I’ll try to sketch a few thoughts on the subject.
The original impetus came from the failure of my electronic style-modeling music to connect with audiences. I had implemented algorithms that built statistical models of a song or style that allowed me to generate original music in that style. These algorithms worked on a symbolic (note-based) level, so the results of the algorithm had to be manifested as audio with synths or samplers. So, I was a guy on stage with a computer, and the music came out of speakers. No one had any idea what I was doing to create the sounds they were hearing, and for all the audience knew, I pushed play and checked my email. Continuing in this vein was a dead end.
My thinking was that if I built physical forms for my artificial musical agents, the interaction between me and the agents would become more understandable, and the performance as a whole would take on more theatrical appeal. Furthermore, the introduction of a sophisticated controller system would allow me to play or improvise with my agents instead of just directing them. Once the project was underway, this improvisatory performance mode took on greater significance, and it’s only now that I’m circling back to working with machine learning and style modeling.
That’s a sketch of the pragmatic origins of the project, but it doesn’t say much about what I’m doing in relation to other music. I’ll try to lay out of few, still-gestating thoughts on that here, but bear in mind that the band is far from complete. There will be more machines, synths, and most likely, more humans (no, not clones of me).
Yesterday a few people on Create Digital Music commented that my machines lose timbral subtlety compared to traditional performance; they notes that a human percussionist can get a much wider variety of sounds out of djembe than my machine can, and that’s true. But I gain a lot, too. I gain improvisatory power. The controller interface and the mechanics enable playing rhythms that would be impossible or difficult for a human. I can play a simple rhythmic pattern on the djembe, loop it, and add another pattern, loop it, and build up a dense, interweaving texture on a single drum. And then I can push a button and have the whole pattern play backwards. Humans can’t do that. The machines and the controller interface, even in the context of a single instrument, introduce a new and broad palette of improvisatory techniques.
A word about timbre: I tried really hard to make the drums imitate the sound of a human hand hitting the drum, but it turns out this is damn near impossible. Some of the drum strokes I get could be mistaken for a human drum stroke, but usually they sound slightly different. I tried a lot of materials: rubber, foam, shoe inserts, shoe inserts wrapped in pig leather wrapped in masking tape, and most failed. What I arrived at for the djembe is rubber stoppers glued to suction cups wrapped in masking tape and then spray-painted with Plasti-Dip. I welcome suggestions for a better solution.
Control of and interaction with the whole band presents its own set of new improvisatory tools, but they exist on a more abstract level than what I sketched above regarding a single instrument. Performance with the whole group fractures and reflects the identity of performer; I might improvise on two instruments simultaneously, doing a sort of call-and-response between the right and left sides of my brain, and then I could loop that interaction and improvise over it with a third instrument while the computer alters what I originally looped. There is a lot of potential here that I haven’t begun to explore (I’m still focused on improving my chops), but I think this is where much of the value of machine music lies. Human-machine interaction disrupts the one-to-one mapping between the sounding instrument and the musician’s intent. That mapping underlies how we process music as a communicative process; when the mapping is disrupted or altered, the listener has to rethink his or her whole framework for processing music as an emotional, communicative experience. There’s an essay to be written here about how this all relates to human identity in an age of digital social networking, but I’ll leave that piece of aesthetic wankery to some unlucky grad student.
Some CDM commenters uncharitably noted issues of self-presentation, to which I’ll respond: Try improvising on two instruments at once while dancing. It’s hard! But the topic is a valid one. One of the reasons I spent many hours trying to make the instruments look beautiful is that I believe musical instruments ought to be beautiful. It’s not a trivial bonus but a vital part of an instrument’s identity. If the saxophone looked like a rubber hose, it would have a vastly different cultural meaning. Contemporary electronic gear is a huge disappointment in this regard. Beyond the instruments themselves, I am very interested in putting together a compelling theatrical experience. That project involves my own (or other musicians’) appearance, gestures, attitude, etc. The motionless, silent laptop artist is my negative model. Maybe I oversold trying to do the opposite in the video, but I think the intention is correct and just needs some polish.
The problem of cultural associations is a massive one in a project like this. Say the word “robot” and you conjure images of either a dystopian future or kitsch. The T-1000 on the hand and Chucky Cheese on the other. I’ve tried to avoid both by rooting my project in the rich history of machine-made music; the band is named after Al-Jazari, who invented the world’s first robot band in the 13th century, and the machines themselves look like antiques from an alternate 19th century. This rootedness in the past also serves to present an image of technology that is more artesanal than industrial. I want the machines and the performers to have unique identities and not look like mass-produced tokens of an oppressive future where humans have been rendered superfluous. My attitude contrasts with that of a band like Kraftwerk, who fetishize dissolving the messiness of individual identity into a logic of conformity (they’re still great, of course). Having other people in the band will help realize my goal and steer the project away from the image of the megalomaniacal, mad-scientist that I think some may perceive. But there are some technical hurdles to overcome before I can do that. Thanks for listening and for your comments. It’s been a pleasure to read them and respond.