Comparing Pitch Spelling Algorithms on a Large Corpus of Tonal Music

This paper focuses on the problem of constructing a reliable pitch spelling algorithm—that is, an algorithm that computes the correct pitch names (e.g., C]4, B[5 etc.) of the notes in a passage of tonal music, when given only the onset-time, MIDI note number and possibly the duration of each note. The author’s ps13 algorithm and the pitch spelling algorithms of Cambouropoulos, Temperley and Longuet-Higgins were run on a corpus of tonal music containing 1.73 million notes. ps13 spelt significantly more of the notes in this corpus correctly than the other algorithms (99.33% correct). However, Temperley’s algorithm spelt significantly more intervals between consecutive notes correctly than the other algorithms (99.45% correct). All the algorithms performed less well on classical music than baroque music. However, ps13 performed more consistently across the various composers and styles than the other algorithms.