Music recommendation systems have come along way since then, steadily adopting user-friendly features that make the system more personalized for all concerned parties. The main objective in this paper is to implement a user-friendly interface connecting individuals with the music system and enhancing their listening experience through consideration of individual emotional demand as well as musical taste for all different users. For the methodology, Haar Cascade Classifier used to detect faces and Conv2D layer, activation function layers like ReLU as well sigmoid where needed after convolutions in all levels followed by Max Pooling with few Batch Normalization if required have been implemented for recognizing expression from facial space. By crossing the boundary between today’s technology and music enjoyment, the system gave users emotion-appropriate playlists based on their facial expressions. Results of the trials demonstrated that, this proposed system can identify user’s music-preference affective states respectively with accurate evaluation.