Infant Identification from Their Cry

Cry is the only means of communication for an infant. Understanding the properties of infant cry is very crucial for establishing a basis for using cry as a tool for pathological diagnosis or possibly identifying infants. In this paper, an attempt is made to identify infant from their cry. The experiments are shown for linear prediction coefficients (LPC), linear prediction cepstral coefficients (LPCC), and Mel frequency cepstral coefficients (MFCC)and Teager energy based MFCC (T-MFCC) as input feature vectors to the polynomial classifier of 2nd and 3rd order approximation. Results show that MFCC performs better than other features. This may be due to the fact that MFCC is designed to mimic human perception process and hence represent the perceptually relevant aspects of short-time infant cry spectrum.