BABYCRYING

Deafness is an important disability in the world. Hearing screening is a usefull aid for detecting the hearing disease on time to avoid the latency in treatment and rehabilitation process for the infants. Conventional methods same as OAE or ABR tests is expensive and need the experts. Sound production mechanism in the infants depends on physical and psychological condition such as pain, hunger, insult and fear. Based on the hearing feedback, Infant’s cry also affect from the deafness. In this work an infant cry signal processing and pattern recognition for deafness screening is presented and developed. Signals are recorded from 24 deaf and 27 normal cases and after pre-processing Mel Frequency Cepstral Coefficients (MFCC) and Linear Prediction Coefficients (LPC) and Cepstrum coefficients of the Linear prediction is used for feature extraction. Same as the conventional screening methods, classification to the “Pass” that means normal and “Refer” that means need to more experiment is done by MLP Neural Network and some post processing. Simulation results show that based on proposed signal processing approach for screening, discernment successfully all of the cases to the “Pass” and “Refer” correctly.