Year of Graduation
Level of Access
Restricted Access Thesis
Department or Program
Music forms the soundtrack to daily life and serves as an important cultural marker for people around the world. As the world becomes digitized and connected via the internet, the opportunity is increasingly accessible for anyone to share music with the world and to create chart-topping music that defines the cultural vernacular. Many prominent producers have little to no formal musical training, especially in Western music theory. As a result, loop-based music dominates the lists of most-played music on the radio and streaming services, often not deviating from basic functional harmony. With this project, I have created a compositional tool in the form of an iOS app which identifies the harmonic “fingerprint” behind a given set of songs. The app then leverages this understanding to create sequences of chords in the style of that fingerprint. To accomplish this, the app employs web scraping to create a corpus of musical information in the form of Markov chains — a transition table which underlies the data set. I introduce the idea of musical “chunks” defining a harmonic “fingerprint” and various methods of traversing the transition table to create chord progressions employing the fingerprint as a guide. The tool allows for specification of corpus, chunk size, traversal method, and the ability to listen to, share, and save generated results. The resulting app is a tool that allows the user to answer the question: “What would happen if Stevie Wonder and Billie Eilish wrote a song together?”
Available only to users on the Bowdoin campus.