Music Recommendation System Based on Facial Expression

Abstract

In the modern digital age, where users face an overwhelming array of music choices, personalized recommendations are crucial for enhancing the listening experience. Studies indicate that music significantly influences emotions, affecting mood, stress levels, and overall well-being. Over 60% of users struggle with decision fatigue when selecting music from their vast collections. To address this challenge, we propose a real-time music recommendation system that utilizes facial expression analysis through deep learning. The system employs a Convolutional Neural Network (CNN) trained on the FER-2013 dataset to analyze facial expressions captured via a webcam. Detected emotions are processed in real time, enabling dynamic music selection. Flask is used for backend API development and server management. Supabase functions as a cloud database. This emotion-driven approach streamlines the music selection process, saving time and reducing the stress of manual browsing.