Emotion Based Music Recommendation System

Abstract

Today, music platforms provide easy access to many types of music. They continue to strive to improve music organization and research to solve the problem of selection and make discovering new music easier. Recommendations have become popular to help people choose the right music for every situation. But there is still a difference when it comes to personal and emotion-focused recommendations. Music is beneficial to humans and is widely used to relax, regulate emotions, eliminate stress and illness, and regulate mental and physical activity. Music therapy has many therapeutic areas and applications to improve health. This article will introduce the design of music recommendations that are guided by the user’s thoughts, feelings, and content of activities. We've crafted a Convolutional Neural Network (CNN) model tailored to recommend music based on the user's facial expressions. which help people choose music according to their different mood. It has Validation Accuracy of 97% and Testing Accuracy of 76%.