Wavelet analysis of newborn infant cry

The acoustical analysis of the infant cry is a non-invasive approach to assist the clinical specialist in the detection of abnormalities in infants with possible neurological disorders. Along with the perceptual analysis, the automatic analysis of the cry is often carried out through commercial or free software tools. However, the neonatal cry is a signal extremely difficult to analyze with standard techniques due to its quasi-stationarity and the very high range of frequencies of interest. To address this issue, we present a new fully automatic method that exploits the wavelets high time-frequency resolution and low computing time properties for the estimation of the fundamental frequency F0 and vocal tract resonance frequencies F1-F3. The method is tested on synthetic signals giving results comparable to existing tools. It is also applied to a set of 1669 newborn cry units (CU) coming from 10 very preterm babies and to a set of 3514 CUs of 20 full-term infants.