Music Through Machine Learning

Building Instruments Using Classification, Regression, & Neural Networks

This semester, under the guidance of Professor Evan Peck at Bucknell University (utilizing Professor Rebecca Fiebrink's online course "Machine Learning for Musicians and Artists" offered through the University of London), I learned how to utilize machine learning in the process of creating new and exciting electronic instruments. There are examples of various different ideas/methods that I worked with throughout the semester in the labeled sections, and on the link above, you can watch a video of one instrument that I trained using a linear regression model that takes in data from 100 pixels on a webcam and produces a variety of different sounds based on what it sees (in the example below, the speakers produce different sounds depending on how much of my stuffed animal, me, and nothing that it sees). I hope to utilize what I learned this semester in both my Computer Science and performance endeavors going forward, and I am excited to see what innovations/learning opportunities that this may bring. Enjoy!