A Multilayered Approach to Automatic Music Generation and Expressive Performance

摘要

When analyzing scores, musicologists often use multilayered representations to describe different importance levels of notes and chords, according to hierarchical musical structures. These structures are believed to represent the composer's mental representation as well as the listeners' perception of the piece. Thus, in the context of automated music generation, this kind of information can be of great use to model both the composition itself and its expressive performance. In this paper one computational method to perform this kind of analysis is described. Its implementation is then used to generate short musical phrases according to a hierarchical structure that is also used to model the performance of these melodies.