Simulink model of Mel Frequency Cepstral Coefficient analysis for extracting asphyxiated infant cry features

Abstract

Pathologic conditions such as deafness, asphyxia and others can be detected from infant cries in real time with a digital signal processor. This is only possible if the digital signal processor is embedded with a Simulink function model for the requisite signal processing. Prior to detection, it is found from previous works that the features of infant cry have to be extracted using the Mel Frequency Cepstral Coefficient analysis. Since the Simulink function model for MFCC analysis is currently unavailable, it then becomes the motivation of this paper to create it to be added to the Simulink function model library. This paper describes the development of the Simulink MFCC analysis model for extracting significant features of asphyxiated infant cry. The model is developed using a combination of four blocks of Matlab M-files. The output from the Simulink MFCC analysis model have been verified against those generated from the Matlab M-file algorithm from our previous work. They have also clearly shown differences between features extracted from normal and asphyxiated infant cry. As such, these features can then be used as input to a classifier. The classification of asphyxia from infant cry can become real time by downloading the Simulink MFCC analysis model and classifier to a Digital Signal Processor board, to become an assistive tool to the clinicians.