Influence in Early Electronic Dance Music: An Audio Content Analysis Investigation

Audio content analysis can assist investigation of musical influence, given a corpus of date-annotated works. We study a number of techniques which illuminate musicological questions on genre and creative influence. By applying machine learning tests and statistical analysis to a database of early EDM tracks, we examine how distinct putatively different musical genres really are, the retrospectively labelled Detroit techno and Chicago house being the core case study. Further, by building predictive models based on works from earlier years, both by a priori assumed genre groups and by individual tracks, we examine questions of influence, and whether Detroit techno really is a sort of electronic future funk, and Chicago house an electronic extension of disco. We discuss the implications and prospects for modeling musical influence.