This music recommendation system that utilizes facial expressions to determine the user’s mood and preferences. The device is made to function both with and without a face mask, making it useful and adaptable in the pandemic situation of today. Although prior studies have looked at music suggestion and facial emotion identification separately, the particular setting of face masks poses a special problem as well previous research did not focus on the accuracy of the models and they did not used multiple emotions. This issues necessitates research into how emotions are perceived and expressed when a mask is partially covering the face through outward facial features like the eyes and brows. In order to accurately recognize facial emotions and offer music, a revolutionary strategy combining cutting-edge methods, specialized convolutional neural network (CNN) layers, and modified activation keys must be developed. Face masks offer a challenge, thus the solution uses cutting-edge methods to evaluate the remaining exposed facial features, like the eyes and brows, in order to efficiently capture emotional information. The researcher suggests modifying current CNN designs to do this by adding specialized layers that concentrate solely on the visible facial regions. The solution tries to improve the network’s capacity to derive meaningful emotional information from the observable face features by customizing the network’s design, for as by including attention mechanisms or feature fusion modules. The researcher also investigates the usage of unique activation keys that direct the network’s focus on pertinent face regions and emotional cues while ignoring masked regions. The technique seeks to enhance the accuracy and robustness of facial emotion recognition by providing customized activation keys that emphasize the visible facial areas and suppress the impact of face masks.