terleski@cmu.edu - resume (pdf/doc) - weblog

For my undergraduate thesis, I wanted to make it possible to explore the musical similarity of songs in an easy way. Using iTunes, a user could create a playlist and, with the click of a button, have a new playlist created based on the similarities found within the songs on the created playlist. This was made possible through musical analysis using Fast-Fourier Transforms, Markov Models, and a k-nearest neighbor algorithm.