Classification and detection of classic Turkish music maqams

Various studies have been carried out from past to present on the classification of Classical Turkish Music maqams. During the classification process, there were many disagreements and different systems were developed. In addition to this, personal mistakes were often experienced in the musical note extraction. This situation creates problems in learning the maqams and recognizing the maqam being listened to. In addition, it poses a challenge not only for those who receive Classical Turkish Music education, but also for individuals who are interested in this music genre. With the development of technology, solutions are sought by using signal processing method with computer-based programs. In this study, a musical information retrieval (MIR) system has been developed to classify the maqam types of songs performed in different maqams in Classical Turkish Music. Time wavelet scattering method was used to extract the features of the audio signals over the system. The songs with feature extraction are classified by machine learning. Graphics processing unit accelerator is used to improve system performance. After the classification process, the maqam type can be determined by uploading the song that the user wants maqam detection to be made into the system. The difficulty in classifying the mentioned Classical Turkish Music maqams has been overcome for the maqams in the study. In the system, feature extraction, classification and detection processes are designed over MATLAB.