Facial Emotion Song Recommender System

Abstract

Sometimes, it is very difficult for someone to determine whether a person wants to hear a particular music from the vast array of available choices. So, this paper has proposed a new concept of playing music that is based on the emotion of the user. The primary goal of the music recommendation system which is proposed in this paper is to offer customers recommendations that match their tastes. The most current view of the paper involves manually playing the jukebox, using wearable computers, or classifying based on auditory characteristics. Understanding the user's present emotional or mental state may result from analysing the user's facial expression and emotions. One area is having a great possibility to provide the audience, with a vast variety of options that are based on their preferences and music and video.

In this paper, the primary goal is to show a playlist of songs on any particular music application (YouTube/Spotify) based on each person’s mood. Several images of the user are collected at that precise moment using a camera with the user's consent. To determine a person's mood, these photos go through a thorough testing and training process. For this, the deep learning technique called CNN is used to categorize various emotions. After this, based on the trained model, the various emotions are categorized and based on this the music playlist is generated.