MOODMUSIC: An Android App for Mood-based Music Player using Deep Facial Expression Detection

With the growing popularity of mobile applications, there has been a surge in the integration of advanced features to accommodate user’s personalized behavior with an indulging experience. The evolution of music player applications to align with the dynamic moods of users necessitates expansions in current music app functionalities. This research work aims to design an optimized music player app that needs to interpret facial expressions to detect mood and play the recommended song based on the mood to rejuvenate the user’s ongoing emotional state. An 8 layer convolutional neural network model is proposed for deep facial detection. The performance achieved on the FER 2013 dataset is around 89% on the training dataset and 60% on test dataset. The accuracy achieved for each mood is 41.1%, 26%, 28%, 82.7%, 58.8%, and 73.9% respectively for angry, disgust, fear, happy, neutral, sad, and surprise mood.