Computational Modeling of Pop Music Structures: Searching for Magical (and Maddening) Musical Moments

Most computational musicology focuses on the analysis of musical data, with an aim to deliver insights about music while most music cognition focuses on the analysis of human-response data and typically aims to deliver insights about human cognition. However, these two disciplines, when combined, offer immense opportunity for joint insights. In this talk, I will illustrate the potential for musical data to help answer questions and test hypotheses about music cognition, drawing on research from two different projects: one in which we model the musical structures that may be driving emotion perception in music; and another that uses human-generated musical data to examine the musical features of an "earworm" and what may be unique about them. Both of these projects involve analyzing popular music; a challenge to computational musicologists relying on symbolic (text-based) musical data as popular music is typically not printed. As such, this talk will also touch on some novel tools and resources for computational musicologists wishing to engage with popular music materials.