Emotion Based Music Player

Emotion Based Music Player The study of music and emotions suggests that there is a psychological relationship between a person’s emotional state and the type of music they listen to. The purpose of this project is to understand and analyze various algorithms for an emotion recognition system. This project is split into two halves; extracting metadata from songs to determine their genre and using machine learning to determine the emotion of the user. The project centers heavily towards machine learning and data mining, where different techniques are used to determine the emotion of the user and analyzes the implications of using each technique. Using the determined emotion, the final music library of the user is sorted out. This report begins with a review of the existing feature extractors and the current techniques used for emotion recognition before discussing the development approach and design of the system. Having completed the application, the performance of each of the different techniques is reviewed. A comparison and analysis of the different feature extractors used is made. These results are then compared against real world results. Finally I discuss further research opportunities which could improve the application. Student: Umer Iftikhar Supervisor: Dr. Tim Morris