Visual Pattern Analysis using Digital Sheet Music

Abstract

The increasing amount of digital sheet music such as MusicXML [5] and through databases like MuseScore1 enables the creation of innovative applications for visual analysis of harmonic patterns. Visual music inspection [2] allows for the investigation of harmonic progressions, the extraction of relevant themes or melody motifs [4], understanding tonal relationships [3, 9] as well as the overall structure of sheet music [6]. This visual inspection, however, requires the ability to read the common music notation [10]. MusicXML preserves both layout information and musical content allowing for accurate digital representations of sheet music and additions of visualization annotation overlays to visually highlight pattern. Many existing approaches consider primarily audio data to extract melody [1] or chords, which requires different models for pattern recognition than digital sheet music.