A System for Detecting of Infants with Pain from Normal Infants Based on Multi-band Spectral Entropy by Infant's Cry Analysis

Infant cry is a multimodal behavior that contains a lot of information about the infant, particularly, information about the health of the infant. In this paper a new feature in infant cry analysis is presented for recognition two groups: infants with pain and normal infants, by Mel frequency multi-band entropy extraction from infant's cry. In signal processing stage we made pre-processing included silence elimination, filtering, pre-emphasizing. After taking Fourier transform, spectral entropy was computed as single feature of signal. In classifying stage, by training artificial neural network, correction rate of recognition was obtained 66.9%. In order to enhancement in results, we used Mel filter bank. Entropy of each sub-band constitutes elements of next feature vector. We used PCA analysis for reducing in dimension of the recent feature vector. After ANN training, correction rate improved to 88.5%. So multiband spectral entropy enhanced results in salient correction rate.