A Content Analysis of the Research Approaches in Music Genre Recognition

Abstract

Music is part of many areas throughout daily life, from entertainment to rehabilitation. Technological developments are widely used in music, as they are in every field. In this context, very large music databases exist both online and offline. The importance of categorizing these databases and classifying them by genre has increased. Manual categorization of music databases has a high margin of error and takes a long time. Although various tools are used for this purpose today, developments in machine learning algorithms have led researchers to work in this field. The present study is a content analysis of the studies conducted in recent years to determine music genres. Features, datasets, and machine learning methods used for recognizing music genres are reported in detail. The main purposes of this study are to describe the features of the databases used and to create an abstract on the effectiveness of the features and machine learning methods used in previous studies. Thus, it is expected to shed light on future research.