Music recommender system via deep learning
Music is an integral part of our everyday life. Almost everyone uses various streaming services to listen to music on their mobile phones. These streaming services recommend tailored music to each individual by categorizing music into numerous genres, thus enriching the user experience. The backbone...
Saved in:
Published in | Journal of information & optimization sciences Vol. 43; no. 5; pp. 1081 - 1088 |
---|---|
Main Authors | , , , |
Format | Journal Article |
Language | English |
Published |
Taylor & Francis
04.07.2022
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | Music is an integral part of our everyday life. Almost everyone uses various streaming services to listen to music on their mobile phones. These streaming services recommend tailored music to each individual by categorizing music into numerous genres, thus enriching the user experience. The backbone of these streaming services is a music recommender system that recommends the best music to the users. The efficiency of these streaming services ultimately depends on the music recommender systems' results. The music recommender system suggests music based on the user's interests. The necessity of these music recommender systems is increasing day by day, but most of the systems do not meet users' expectations. This paper plans and actualizes a general music recommender system that applies several audio signal features to train several machine learning and deep learning models for categorizing music into ten genres to meet the users' expectations for music suggestions. The music recommender system consists of a feature extractor, a data preprocessing module, and a classifier. The system classifies the audio signals into blues, classical, country, disco, hip-hop, jazz, metal, pop, reggae, and rock. The classifiers such as support vector machine (SVM), K-nearest neighbors (KNN), random forest, and feedforward neural networks (FNN) are used for the system evaluation. The music classification results from our recommender show that the FNN based system has the highest prediction results than other classifiers. The FNN based music recommender system achieved 80% accuracy on the GTZAN music dataset and validates the impact of the deep learning models for music classification. |
---|---|
ISSN: | 0252-2667 2169-0103 |
DOI: | 10.1080/02522667.2022.2094546 |