The Common Cold: Using Data Science to Define the Winter Topic in Video Game Music. This presentation draws on my collaborative work with Evan Williams, a music-theorist-turned-data-scientist. It models a new approach to theorizing topics via music informatics. Whereas other topic theory research typically relies on the author to recognize connections among musical features (Agawu 1991, Monelle 2006, Atkinson 2019), we allow the data to suggest its own groupings, revealing relationships that may not otherwise be apparent. Our case study is the winter topic. We chose to focus specifically on winter in video game music, as video game music leaves little ambiguity around what the music ought to signify. Video games commonly have an icy or snowy area, complete with cold-weather creatures, landscapes, game mechanics, and music for the player to encounter. Our dataset has over 160 examples of such music, representing games on all mainstream platforms (Nintendo, PlayStation, computer, etc.) and spanning the years 1987–2020. Each example is tagged with its musical features: instrumentation, meter, tonality, presence/absence of arpeggiated accompaniment, amount of reverb, and drum pattern. We use Python, the PyData stack, and standard data science algorithms like PageRank alongside traditional music analytical techniques to illuminate several facets of the winter topic. Through this case study of winter video game music, we present a model of analysis that could easily be adapted to suit any repertoire or topic.
Megan Lavengood is Assistant Professor and Director of Music Theory at George Mason University. Her research primarily deals with popular music, timbre, synthesizers, and recording techniques. Her current research project focuses on topic theory and video game music. Before COVID, she was a soprano in a Renaissance quartet.