EEG based detection of speaker familiarity

In forensic sciences, finding the connection between members of a criminal organization is quite an important task. The regular and organized crimes which are committed by a group of people are accepted more dangerous than the crimes which individuals did. Detecting the members and solving the hierarchical structures of these type organizations are accepted as the first step to define the organization and catch the members. The members of these type organizations which was caught by security units could be in an effort to keep the secrets of the organizations. The purpose of this master thesis which is offered as an assistive application to this task is finding whether a connection of a member with another member of an organization is existed. This thesis is presented as an alternative way to assist this task. The studies which shows that the speaker familiarity could be built by a long and closed relationship, led us to choose the thesis topic. As pioneer application in the literature, whether someone could recognize a person from his voice or not, could be detected with the mean accuracy of %80,5 and the standard deviation of 3,7 on EEG recorded from scalp of six subjects.