Enhancing Music Recommendations Using Machine Learning

A substantial amount of the mainstream music recommendation systems are noticed to shine among well-known artists while leaving independent artists without visibility. This study aims at developing a fairer recommendation system promoting independent music by re-ranking songs from underrated artists in the search results. Data is collected from the Spotify API for the proposed recommendation system, which will include general popularity metrics and genre types. Also included will be preprocessing actions like genre encoding and popularity normalizing. A Random Forest classifier is trained and optimized using GridSearchCV for accuracy improvement, together with precision, recall, and F1 score. The final model achieved an overall accuracy of 93 %, an 89 % precision, 93 % recall, and an 86 % F1-score. it has been deployed in the real-time Streamlit-based web application for user-centric recommendations. Hence, this can be a possibility in place of traditional algorithms while promoting music diversity. Future improvements could use deeper audio analysis using tempo, rhythm, and pattern features, as well as user history and preferences as inputs through deep learning for better personalization and engagement.