In a study, recently published in Music & Science, DCML researchers Fabian C. Moss, Markus Neuwirth, and Martin Rohrmeier use Latent Dirichlet Allocattion (LDA), one of the most pervasive topic models in the Digital Humanities, to uncover underlying “tonal profiles”: distributions of pitch classes. Tracing topics over ~600 years shows historical changes in tonality, in particular at the transition from common-practice to extended tonality.