Enhancing Music Discovery: A Real-Time Recommendation System using Sentiment Analysis and Emotional Matching with Spotify Integration
J A Carol Jeffri, A Tamizhselvi · 2024
This music recommendation system is designed to suggest songs based on the emotional state of the user, utilizing advanced Natural Language Processing (NLP) and sentiment analysis methods. It employs Hugging Face’s aiknowyou/it-emotion-analyzer for analyzing both text and audio inputs, alongside pysentimiento for sentiment classification. Detected emotions are mapped to specific music genres, with song recommendations retrieved using the Spotify API. Preferences for specific artists can also be set, enabling personalized song suggestions that consider both emotional analysis and artist preferences. While text-based emotion detection has shown high accuracy, audio-based recognition poses challenges due to complexities in processing sound. The user interface, built with Streamlit, offers an intuitive and efficient platform for engaging with the recommendation system. Future enhancements may focus on improving audio signal processing, refining the emotion-to-genre mapping, and incorporating reinforcement learning to further enhance adaptability and personalization.