Music recommender system based on emotions from facial expression

Abstract

Music has long been recognized as a powerful medium for emotional expression and regulation. This project presents EmoTunes, a desktop application that integrates facial emotion recognition with Spotify’s recommendation system to deliver personalised music playlists tailored to the user’s current emotional state. By leveraging Amazon Rekognition for real-time facial emotion analysis, the system identifies eight core emotional states, which are then mapped to corresponding music playlists. Users can either link their own Spotify playlists or generate new ones, pre-populated with songs from Spotify’s curated emotion-themed mixes such as “Happy Mix” or “Sad Mix.” The integration of machine learning and natural language processing in Spotify’s music classification enhances the relevance of the recommendations. The system aims to not only provide music for entertainment but also improve users' emotional well-being through responsive and emotionally intelligent playlist generation. Future enhancements include the integration of voice tone analysis and user feedback loops to improve emotion detection accuracy and create a more adaptive experience.