Chord Based Feature Engineering for Genre Classification in Popular Brazilian Music

Abstract

We propose techniques for the extraction of both chords, through web scraping, and their features, and use it to find what are the most important ones for the characterization and classification of popular brazilian music. The features where extracted based on harmonic grounds. The classification was made using the Random Forest algorithm. Final results show that brazilian music can be characterized by its harmonic structures.

Date
Event
Undergraduate Thesis
Location
Curitiba, ParanĂ¡, Brazil
Links