Developing an artificial intelligence based gender detection model for forensic voice analysis

Today, with the development of technology, we come across sound data in many different areas around us. Speaker recognition systems will facilitate the detection of crimes committed with voice data. Many features such as age, gender and mood are detected by speaker recognition systems. These detected features are important features in a speech. Knowing the identity of the speaker is useful in different areas such as the voice response systems we use today and forensic applications. It is also seen in legal events in the field of forensic informatics. Various features of the speaker's voice were examined and the studies to determine the age and gender of the speaker were examined with the clues to be extracted in line with these features. Different properties of sound are explained, information about signal structure and analysis is given, sound forensics is mentioned. The classification methods that can be used are compared. With the collected data set, a model has been developed that can be used to determine gender from the speaker's voice. In the model, a one-dimensional three-dimensional local pattern was used for feature extraction and SVM was used as a classification method. The accuracy rate of the proposed method was calculated as 99.59%.