A System for Identifying Common Melodic Phrases in the Masses of Palestrina

Abstract

An important direction in Renaissance music research is the relationship of repetition to structure and compositional practice. This calls for very time-consuming analysis, and is generally unfeasible when applied to large collections of compositions. However, such large-scale analyses are necessary to allow us to identify trends across entire repertories. We describe a system based on the use of suffix arrays, which allows us to find instances of melodic repetition in an enormous body of music, within a reasonable amount of processing time. This system identifies all transpositions, inversions, retrogrades, and retrograde inversions of unknown melodic segments. It has been applied to the entire collection of masses by Palestrina, a corpus of over seven hundred mass sections and approximately one million notes.

Notes

¹Palestrina's masses were made available in electronic format (Humdrum encoding) in 2004 by Bret Aarden, currently at the University of Massachusetts—Amherst. This electronic edition, originally produced by John Miller in 1992, is based on the Casimiri edition of Palestrina's complete works (Palestrina 1981).

²Each of these duos of course also contains a match within itself, due to imitation at the fifth.