Genre and mood classification using lyric features

Abstract

Musical genre and mood classification techniques combining lyric and audio features for Music Information Retrieval (MIR) have been studied widely in recent years. This paper investigates the performances of musical genre and mood classification using only lyric features. In this preliminary study, the Part-of-Speech (POS) feature is utilizes for classification of a collection of 600 songs. Ten musical genre and mood categories were selected respectively based on a summary from the literature. Experiments show that classification accuracies for mood categories outperform genres.