Identification of Infants' Cry Motivation Using Spectrograms

Understanding the motivation of adults and children expressions during different sensations as hunger, pain, etc., is not a hard task. This is due to their ability to communicate their feelings using sound expressions that are commonly used among human beings and are easily understood by them. Infants, on the other hand, lack the knowledge to show their feelings, which makes them express themselves through crying and by making facial expressions that can differ depending on the baby's age or even on the birth condition. Having that in mind, this paper aims to automatically identify whether an infant's cry is motivated by the feeling of pain. To accomplish this goal, features extracted from the spectrogram generated from audio samples of crying babies are classified using a novel dataset, which is carefully designed, curated and made available to the research community for the development of classifiers in this field of research.