An evaluation of human perception for neonatal cry using a database of cry and underlying cause

This study aims to test the hypothesis that whether caretakers are able to perceive the basic needs of neonates solely based on cry. A database of 315 neonatal cries was built and systematically labeled in postnatal nursery at a government run tertiary hospital at Kolkata. All the cries were labeled as hunger, pain, wet diaper and others constituting 4 different classes. A computer based evaluation system generated fifty or more cries in a random order for participants to listen. Participants were expected to predict the causes of cries after a preliminary training session. A significant proportion of participants were able to classify them into reason based classes. There is a fair agreement in these interpretations among evaluators. Results support our hypothesis. The study indicates the existence of a perceptible component in the cry signal. Auditory feature variations in the neonatal cry are perceptible by the human ears guiding an individual towards an effective need based classification. The study emphasizes the possibilities that adults can be trained to correctly perceive the needs of a baby by using a cry database. Such training systems will definitely improving the care of newborn.