In 2001, when I was in high school, I applied to Stanford University. I was truly excited about the topic of algorithmic music, and I tried hard to express that in my application. In the US, for some colleges, the essay can matter a lot. Perhaps that's one of the reasons why I got into Stanford and not UC Berkeley. I was thinking hard about majoring in either Music or Computer Science (but ended up majoring in Management Science & Engineering, since I was interested in too many topics).
It's funny that I was deeply inspired by computational creativity then, and now, 20 years later, I am obsessed with it again. (I think AI art is a subcategory of computational creativity, loosely speaking.)
It's also amazing how far we've come technology-wise since 2001. I took Stanford's Intro to AI then as a high school student, in the middle of an AI winter, and my conclusion was basically "AI doesn't work." The recent explosion in visual AI art is nothing short of spectacular, and, although less prominent on social media, people are using deep learning methods like GANs and VAEs (variational autoencoders) in interesting ways to generate good music as well (check out the videos/papers from the recent AI Music Creativity Conference, especially the keynote on performing music with AI by Nao Tokui). The rate of improvement is amazing.
〜〜 below is the 2001 essay (warning: it is a bit corny/cringeworthy) 〜〜
Six months ago, Serendipity hit me, picked me up, and threw me towards the sky at escape velocity. It was in the form of Professor Diana Dabby (Olin College), who made an extraordinary musical presentation at my high school. All the students filed into the gym and sat on rows of green chairs, awaiting boredom. Still sweaty from basketball, I fanned myself with the book I was going to read during the assembly; I wanted to make good use of assembly time. But the subsequent show would turn out to be worth a thousand books.
First, the professor played a tape of an excerpt from J.S. Bach’s A Well-Tempered Clavier, a familiar Baroque piano piece. She then played a strange, yet euphonic modification of the selection. After its oddly bittersweet conclusion, Professor Dabby calmly revealed the genius behind the modified composition: a computer. A computer!? “That’s right,” she affirmed, “I just used a computer to combine the song and some mathematical equations.” While keeping the essence of Bach, the computer had brought something entirely new out of the piece, infusing an arbitrary aspect into it. Reminiscent of my past contemplation, the dreamlike idea of the integration of math and music was now right in front of me.
It got better, aurally. After a few sweet minutes of Bach, the professor proceeded to play the original and modified versions of George Gershwin’s “Prelude No. 1,” a Contemporary piano piece that I had recently mastered. Her choice of “my song” made me realize and appreciate the universality of my music education; like math, music was really a worldwide language. The professor’s presentation immersed me in that language, urging me to harness its true power. My hands clenched a few times, enjoying their newfound importance and potential. Before this experience, improvising here and there, exhibiting a jazzy, rubato style, I had believed that my performance of the prelude had reached its musical limits. But for me, listening to and swallowing the Gershwin “remix” – with its increased jazziness, chaos, and uniqueness – transcended any previous experience of the piece.
After the prelude was over, my rapture was not; Professor Dabby started to explain the music’s mathematical mechanisms. She reached into her briefcase, produced a transparency, and placed it on an overhead projector. I saw a 3D graph, but it looked more like tangled thread than anything else. Pointing at it, Dabby explained that the three coordinates of each point corresponded to the pitch, amplitude, and length of a note. To put the song into the computer, the professor had mapped each note of the music to a point on the graph. She could then change the coefficients in the equations to transpose the notes to new positions and thus create a different song. After experimenting with the various modifications, Dabby could discriminate among them. The result: a collection of hundreds of compositional changes no human could even think of, done with the click of a mouse and a little editing. A masterpiece. Although the method was a little recondite, I felt it was both nothing I couldn’t do and the beginning of what I could do.
When her presentation was over, the professor lingered in the gym. I approached her eagerly, somewhat like a youth approaching a kung-fu master for acceptance as a disciple. Despite my sweaty clothes, she greeted me warmly and tried to satiate the curiosity in my plethora of questions. She smiled at me, “With this type of thing, there’s a whole lot you can do.” “I know,” I thought, “’a whole lot’ is exactly what I aspire to do.” I thanked her and left the gym with a wondrously amplified respect for the technological world.
I whispered to my friend, “Wouldn’t it be CRR-A-AZY if she used the computer to mix Bach and Gershwin?” She nodded in simultaneous agreement and doubt, a doubt vanquished by Dabby’s climactic composition, the affectionately named “Gershbach.” I had downloaded many techno remixes of classical songs, but the Gershbach was far beyond any conventional mixing. I fell in love.
The next day in my calculus class, the teacher, a fervent lover of classical music, explained the Lorenz system, or the tangled thread Professor Dabby had displayed. The system was part of chaos theory, formally defined as the qualitative study of unstable aperiodic behavior in deterministic nonlinear dynamical systems; in other words, the study of patterns in seemingly patternless occurrences. The Lorenz system had been modeled on the movement of a bucket-waterwheel with holes in the buckets and water pouring from the top. Every so often, the wheel would change direction due to the imbalance caused by the holes. After hearing this, I opened my mouth and kept it that way for an unusual amount of time. I pictured the tangled thread and waterwheel. They looked quite weird and didn’t seem related at all. Man had created music from a waterwheel. If this was true, why couldn’t I create music from, say, the aimless, buzzing fly on my desk? Since then, I respected math even more and became determined to master it, so that I might one day create something crazy and useful like the Lorenz system.
Really, this was all just the beginning. Perhaps my excitement was naïve. Perhaps this math-music concept was old stuff. But no matter how much had been done on the subject, there would be infinitely more to do…infinitely more that I could do. I thought to myself, “Isn’t math a part of everything, including iguanas? Isn’t music a part of everything, including fire? Isn’t everything a part of everything?” I developed an ambition to create much with mathematics, and much with music. For example, I could use math to create visual art or attach electrodes to my head to monitor my emotions and create music from the graphical representation of my sensations. Heck, these kinds of activities became my goals in life. And they were just the beginning.