Reproduce “Computational Features for Symbolic Melody Analysis” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
David M. Whyatt, Peter M. C. Harrison
Researchers compiled 282 melodic features into a structured taxonomy and developed `melody-features`, an open-source Python package for their extraction. The package achieved 0.9920 accuracy in classifying European vs. Chinese folk melodies and an interpretable 8-factor solution maintained 0.9256 accuracy, enhancing accessibility for computational musicology and music psychology.
Key insights
- A comprehensive and unified set of melodic features can achieve very high accuracy (0.9920) in distinguishing musical styles, as shown in classifying European vs. Chinese folk melodies.
- Exploratory Factor Analysis can condense the feature set into an 8-factor solution, maintaining strong classification performance (0.9256 accuracy) while significantly improving interpretability for music psychology applications.
- Pitch-based features (e.g., intervals, contours, tonality) contributed most significantly to the classification of melodic styles, outperforming rhythmic and combined features.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.