Babies’ voices: A collaborative research program on the automated acoustical analysis of the preterm newborn cry

The aim of this collaborative work is to provide the automated assessment of the melodic shape of the newborn cry with the BioVoice software tool. The method was tested on synthetic signals with 100% matching. Acoustical parameters of cries obtained from preterm and term newborns in Liege (Belgium) and Firenze (Italy) were estimated with BioVoice. The automated classification was first compared to the perceptual (visual) analysis considered as the gold standard on a set of healthy at term newborns with a matching up to 85%. Then, significant differences were found between at term and preterm babies up to 85%. Our study suggests that some melodic characteristics of the newborn cry could be detected to predict the belonging to term/preterm group of patients with an acceptable accuracy.