Presenting a set of computer analysis programs that identifies repetitions and studies phrase structures in the annotated corpus of Mozart piano sonatas

Abstract

The advancement of computer technology has the potential to revolutionize the study of music. Western Classical music which is based on notation, has the particular advantage of score analysis. Similar to natural language processing, the building of corpus is foundational for computer analysis. The DCML lab at EPFL has been curating a corpus project. Hentschel published a corpus of annotated Mozart piano sonatas in 2021 which includes annotations that delineate phrases. This paper builds on the phrase annotations of Hentschel and studies the structure of all the phrases. It identifies repetition as an important marker of phrase structure, presents methods to identify, represent, and visualize repetitions, and applies the methods on the corpus of Mozart piano sonatas. Confirming observations from music theory, phrase types including the sentence and the period constitute a significant percentage of all the phrases. There are nonetheless other constructs such as ABBC, AAAC phrase structures. The organization of each phrase is also related to the length of the phrase.