Cry Signal Analysis to Distinguish Deaf Infants fromNormal Ones

Crying is one of the most noticeable behaviors of infancy. Infant cry signals can be used toidentify physical or psychological status of an infant. Recently, acoustic analysis of infant crysignal has shown promising results and it has been proven to be an excellent tool to investigatethe pathological status of an infant. This paper evaluates different conditions made byMel Frequency Cepstral Coefficients (MFCCs). In feature extraction step of infant cry recognitionwe examined the number of MFC coefficients and first and second derivatives effectson the classification result. For classifying normal and deaf cry signals a Support Vector Machine(SVM) is employed as the classifier. As dimension reduction we computed average ofMFCCs on frames other than silence ones. We developed two methods and compared them.We could achieve the best classification accuracy of 98.52% in method I by 7 MFCC withdelta MFCC which makes a feature vector with 14 elements and 95.54% in method II by 9 MFCC with delta MFCC which makes a feature vector with 18 elements

Keywords:

infant cry , MFCC , SVM , classification , recognition , segmentation ,