Music Recommendation System and Mood Tweak Using Facial Emotion Recognition

Abstract

A user’s emotions can be detected through their facial expressions, these characteristics can be parsed, and the information can be used to improve the user’s experience. With the computing capabilities of on-board graphic processing, there is a new trend of functionality being considered in the field of emotion detection. The use of Convolutional and Recurring Neural Networks (CNN & RNN) to conduct a comparison on which deep learning can be implemented for emotional recognition. Both neural networks are trained employing a dataset with differing types of emotion classes. The trained models are applied employing a music player supporting one’s facial expressions. Within the proposed system, an emotion-based music player is developed to perform real time mood detection which recommends songs as per the user’s current mood. If the user is not satisfied with his mood, he can change it. This becomes a further feature to the normal music player. Customer satisfaction is a significant benefit of implementing mood detection. The goal of this technique is to evaluate the user's image, anticipate the user's expression, and propose a song that is suited for the detected mood in order to provide the best experience possible.