On Wisconsin, Winter 2003, Arts and Culture column [original]
Three centuries ago, Johann Sebastian Bach created masterpieces of musical genius using technology no more advanced than a quill pen. Yet, as three UW-Madison undergraduates have set out to prove, the composer may have been totally PC-compatible.
Jerry Hui, Eugene Chan, and Ryan Wong have spent nearly a year designing computer programs that they believe can learn, and eventually even imitate, Bach’s music. Begun as a class assignment, the project has turned into an exploration of the power of computers and the very nature of creativity.
“If we could be successful in synthesizing music that can fake some people into thinking that it’s Bach, that would show that certain styles of music can be explained by math,” says Hui, a dual major in computer science and music composition who has written music since high school. “We might be able to understand more about the creative process behind music.”
The system uses statistical probability to create a master list of Bach’s go-to devices: how often he returns to a theme, inverts it, or transposes it a few notes. Within a year, the students hope to feed in enough data so that the computer can churn out compositions that mirror Bach’s tendencies — music that Hui expects to sound “like Bach on a bad day.”
Bach’s use of formalized themes and structures has made him the grand nerd of composers, a man studied as much for the inherent logic behind his music as its creative grace. And this isn’t the first attempt to use artificial intelligence to replicate his work. A few researchers have tried to get computers to fill a gap in Bach’s final, unfinished fugue, but no one has yielded anything that doesn’t stick out as decidedly un-Bach.
Hui, Chan, and Wong don’t expect to solve that problem. The real value of their project may be instructional, both by creating and deploying a tool to break down and understand elements of a composer’s style. Just building the foundation of their system was enlightening — they had to teach their programs what notes were, and how notes combined into melodies and harmonies, and it wasn’t just the computer picking up new things.
“I didn’t know a lot about music when we started,” says Wong, a computer engineer. “But I’ve developed more and more interest in it.”
— Michael Penn