Music Genre Detection Using Deep Learning

  • T Sai Ganesh, D Dhanunjai, K Guna Pragna, Gopichand G

Abstract

Now-a-days, many songs get added to the database, each song of different origin - differing in genres
they are tagged by the natives. Therefore, having an algorithm that can tag songs with genres that are
common everywhere without redundancy for example metal and metallic, will help music players to better
recommend songs/music to an individual. In this paper we will develop such an algorithm using deep
learning. The same algorithm can also be used to create playlists depending on the mood, for example
chill, relax, disco. Our algorithm converts audio to spectrograms and uses 1 dimensional convolutional
neural network for learning the features from the songs. The model has given an accuracy of 97% on the
dataset “MangaTagATune” upon validation, thus showing the reliability of the model developed.

Published
2020-05-20
How to Cite
T Sai Ganesh, D Dhanunjai, K Guna Pragna, Gopichand G. (2020). Music Genre Detection Using Deep Learning. International Journal of Advanced Science and Technology, 29(7), 2403-2410. Retrieved from http://sersc.org/journals/index.php/IJAST/article/view/17979
Section
Articles