Emotion-Based Song Recommendation System

Abstract

In current scenario, many music applications are competing with each other to increase their customer base. The user chooses the app based on criteria like less latency, UI friendly, and so on. Existing song recommendation systems suggest songs based on the user’s previous music preferences, such as by examining at his previous song choices, the amount of time he spends listening to music, etc. But classifying the data and preparing separate playlists based on user’s history are time-consuming. But, song recommendation based on user emotions is another advancement where the application recommends songs based on user’s facial expressions. The user emotions can be determined by their facial expressions. Therefore, the need to read a person’s emotions and recommending songs accordingly is something new to explore. Thus, in this paper, we propose a machine learning approach that focuses on detecting human emotions from the input image and creating a music playlist based on the emotions detected. The machine learning model, Convolutional Neural Network (CNN) is used for image classification in this approach. The main goals of the suggested work are to analyze the user’s image, categorize their emotion, and offer songs depending on that feeling to increase customer satisfaction.