Music Recommend System Using Facial Emotion Recognition

ABSTRACT: The system plays music based on detected facial expressions through an automatic process. The method deploys facial emotion recognition instead of standard manual song selection or wearable devices or audio-based classification systems to recommend music. Convolutional neural networks, or CNNs, have been applied to as the central component in our system to recognize emotions that people express through their faces. The system operates by playing automatic music through a custom interface which uses combination of Pygame and Tkinter programming. The efficiency design of our system enables faster processing speed and reduced system expenses. Through this method the system achieves both efficiency and precision. Our model received evaluation through testing against the emotion labelling (2013) dataset because it is a common standard norm for study on facial expression analysis. The built-in camera records facial expressions that get processed to detect emotions from a list of happiness, anger, sadness, surprise and neutrality. The system retrieves music content either from individual playlists or from pre-selected mood-based playlists depending on the emotion detection result. By doing so users no longer require manual effort to organize songs according to their mood-based categories. Through modern computer vision technology integration with basic tools we created an emotional music application which delivers personalizable and fluid listening sessions. Our system provides a smoother music listening experience because it understands users’ emotions as they occur. Key words: Convolutional neural network, pygame, tkinter, music, player, camera, facial detection, sentiment analysis, euphony suggestion, utterances, and pattern recognition