Research Project Title:
Generative Music with Machine Learning
abstract:The main focus of this SuperUROP will be on learning how to produce long term music structure in western classical music. The goal is to make the jump from individually composed phrases, melodies, etc. to fully formed pieces of music with a beginning/ending and overall cohesiveness. The work will likely involve building off of existing work in using variational autoencoders to produce semantically meaningful latent representations for music. We will try various models data representations to evaluate what works best, and determine why some models work better than others.
I'm concentrating in music and I'm involved in the dance community at MIT. So I'm excited about participating in SuperUROP because I have the opportunity to get in depth research experience while also applying my passion for musical arts. Hoping to make something that’s fun to listen to!