Özet:

Music is a sound signal consisting of a variety of complex components that change according to time and frequency. Music is widely recognized in literature in which a variety of emotions raises the listener. The fact that a person is saying that the music he listenes contains a sad or happy feeling may not reveal the feeling he actually feels. But according to the feeling that is felt at the moment of listening to music, electrical volatilities that occur within the brain can more accurately reveal the structure of the real feeling perceived. Identification of human emotions using brain signals has been the subject of current research in many areas. In this study, the problem of recognizing human emotions was addressed while listening to music pieces. Turkish music pieces of different kinds were listened to the participants and studied the electrical waves formed in their brains and tried to recognize happy, sad, relaxing and stressful emotions. Participants were asked to listen to music pieces of different kinds in a noise-free environment. For the classification of emotions, the Electroencephalography (EEG) signals were first taken from different channels, and certain properties were extracted through these obtained signals. The extracted properties are classified using support vector machines (DVM), K nearest neighbor (KNN) and artificial nerve networks (YSA) machine learning algorithms. The best accuracy ratio of algorithms used to train the data set and classify human emotions is obtained with YSA. According to the findings obtained, the method used has been observed to show good performance.