A Comparative Study between Artificial Intelligence Techniques in an Automatic Infant's Pain Cry Identification System

Abstract

This paper presents a comparative study between Continuous Density Hidden Markov Model (CDHMM) and Artificial Neural Network (ANN) on an automatic infant’s cries classification system which main task is to classify and differentiate between pain and non-pain cries belonging to infants. In this study, Mel Frequency Cepstral Coefficient (MFCC) and Linear Prediction Cepstral Coefficients (LPCC) are extracted from the audio samples of infant’s cries and are fed into the classification modules. Two well-known recognition engines, ANN and CDHMM, are conducted and compared. The ANN system (a feed-forward multilayer perceptron network with back-propagation using scaled conjugate gradient learning algorithm) is applied. The novel continuous Hidden Markov Model classification system is trained based on Baum –Welch algorithm on a pair of local feature vectors. After optimizing system’s parameters by performing some preliminary experiments, CDHMM gives the best identification rate at 96.1%, which is much better than 79% of ANN whereby in general the system that are based on MFCC features performs better than the one that utilizes LPCC features.