MUSIC RECOMMENDATION SYSTEM

A actual-time framework has been developed to interpret emotional states from facial expressions and suggest music that aligns with those emotions. The machine procedures live video input through an green face detection pipeline that isolates the place of interest, accompanied via a lightweight convolutional neural community (CNN) trained at the FER2013 dataset to categorise emotional expressions. The model is optimized to preserve dependable accuracy even as ultimate computationally efficient for popular hardware. A threaded seize mechanism guarantees clean and uninterrupted video streaming, while a Flask-based totally web interface delivers an interactive experience that combines live video with mechanically curated playlists. Experimental observations confirmed steady reputation of not unusual emotional expressions—which includes happiness, disappointment, and neutrality—and on the spot retrieval of applicable music recommendations. The proposed framework demonstrates robust capacity for emotion-aware multimedia structures and establishes a reliable foundation for in addition improvements in affect-driven personalization.