Abstract:

As music streaming services have grown popular in recent years, so as music recommendation systems too. Yet, the majority of these systems take user's listening habits and preferences into account, which might not necessarily be an accurate reflection of their present emotion or mood. In this paper, we offered a unique method for music and song recommendation that relies on a Convolutional Neural Network(CNN) to identify facial emotion. Our method uses a real-time video feed of the user's face to determine important facial traits using facial detection algorithms. Following that, these features are fed into a CNN that has been trained using a collection of musical songs and the emotional labels assigned to them. A playlist of music and songs that are appropriate for the user's present emotional state is created using the CNN's expected emotional state. Our findings show that facial emotion recognition is a potent tool for enhancing the precision and customization of music recommendation systems. Also, our strategy may be incorporated into current music streaming services to improve the customer experience

Description:

music recommendation, songs recommendation, facial emotion detection, convolutional neural network, machine learning