A System for Computational Analysis of Music
Gred Cipriano
Committee: Gene Luks
Honors Bachelors Thesis(Jun 2000)
Keywords: Computational Analysis, Music

Note-based analysis of music can unlock the mysterious of why a piece sounds like it does, but is a time consuming process requiring a skilled analyst. While computers will never supplant the creativity required in such a field, they can perform many of the more tedious tasks much faster than a human. Two such tasks are key (or tonal-center) discovery and chord elucidation, or the process of finding chord structures in a song. Presented here are algorithms for both, along with descriptions of their relative strengths and weaknesses. Also included is a program called "MIDIStat" written in Java Swing, that implements these algorithms and provides various other statistics on a given MIDI composition.