Generators

Making a music generation engine that could learn different styles from music scores and recordings was my senior tech project in high school. The thought was to get started on something like LucasArts' iMUSE engine or Sid Meier's Bach generator, only shooting for what unites all music. I... didn't get very far. At all. Tried to learn Lisp, studied fractals and neural networks, dipped into papers on other music generators, outlined code that never got to the point of compiling. Getting nowhere was an issue at graduation time. There was even a question about whether I would, and it meant I didn't get a fancy emblem on my diploma.

It was a great excuse to learn music, though. As the year began, I signed up for the available semester-long theory class and convinced my dad to buy me a cheap guitar, which I gradually taught myself to play. And I've been fascinated ever since. Besides failing, which can be a healthy thing—it's better to take on an interesting challenge and fail than not challenge yourself—I realized that I like improvising my own melodies way too much to want a computer to do that for me. Sour grapes? Maybe. Maybe—but I really think they're sweet. Anyway, neural network research keeps closing in on what I over-ambitiously tried to do in 1999-2000. Eerie stuff!