Symbolic Music Genre Classification Based on Note Pitch and Duration

Abstract

This paper presents a music genre classification system that relies on note pitch and duration features, derived from their respective histograms. Feature histograms provide a simple but yet effective classifier for the purposes of genre classification in intra-classical genres such as sonatas, fugues, mazurkas, etc. Detailed experimental results illustrate the significant performance gains due to the proposed features, compared to existing baseline features.

This research is supported by the $\mathit{HPAK\Lambda EITO \Sigma}$ and $\mathit{\Pi Y\Theta A\Gamma OPA\Sigma\ II}$ national programs funded by $\mathit{E\Pi EAEK}$.