Music Genre Classification Based on Paraconsistency

Abstract

This work aims to build an intelligent system for Music Genre Classification (MGC) based on Para-consistent Annotated Logic (LPA). Using a database composed of representative samples of songs from different styles, namely jazz, bolero, bossa nova, forro, salsa, and swing, as well as a paraconsistent discriminative classifier, a supervised approach was developed to solve the problem. Particularly, the proposed system does not necessarily match songs and genres definitely. Instead, it quantifies the degree of similarity between them.