Melodic Pattern Anchoring for Score Following Using Score Analysis

Building on our previous work in score following, we suggest that research on musical pattern significance, representation and categorization can be usefully integrated into a score follower to automatically identify unique melodic signatures in a composition. These signatures may then be calculated and analyzed over the entirety of a composition providing anchor points. Anchor points replace what has been, in practice, an arbitrary segmentation of scores with a unique division of a composition based on information that is essential to the operation of a score follower. The machine’s understanding of the entire score is enhanced and our algorithm’s performance is refined.