Harmonic Stylometry of Classical Composers

Stylometry is the statistical fingerprinting of authorship — the tools that can tell Shakespeare from Marlowe by counting function words. This project asks whether the same idea works on music: can harmony alone fingerprint a composer?

Working from symbolic scores in the Kern format, we wrote parsers to extract five harmonic features — consonance, dissonance, modal interchange, subdominant borrowing, and “bluesiness” — and turned each composition into a feature vector. Comparing the distributions across Bach, Beethoven, Chopin, Joplin, and Vivaldi surfaces genuine stylistic separation (Beethoven’s consonance, for instance, runs higher and tighter than Chopin’s) that lines up with the genres each composer worked in.

Built in R with a functional-programming pipeline (purrr, ggplot2). A two-person project that let me point quantitative tools at something I care about outside of math — it grew out of the same interest in music that I still build side projects around.