Infants can communicate their internal state (such as pain, hunger, fear, fatigue, or stress) by the nature of their crying. Experts in linguistics suggest that the cry comprises the first speech manifestations. This article describes the design methodology for classifying baby crying sound events according to the pathological status of the infant. Such an automated system can be an aid to an attending physician performing a diagnosis. In order to address this challenge, a great variety of audio parameters (Perceptual Linear Prediction, Mel Frequency Cepstral Coefficients, Perceptual Wavelet Packets, Teager Energy Operator, Temporal Modulation) were considered. Classification techniques, including Multilayer Perception, Support Vector Machine, Random Forest, Reservoir Network, Gaussian Mixture model, and Hidden Markov model were customized. The goal is to provide an automatic and noninvasive framework for monitoring infants and helping inexperienced/trainee pediatricians, parents, and baby caregivers to identify the baby’s pathological status.
Affiliation: Politecnico di Milano, Department of Electronics, Information and Bioengineering, Milano, Italy
JAES Volume 63 Issue 5 pp. 358-369; May 2015
Publication Date: May 22, 2015
No AES members have commented on this paper yet.
If you are not yet an AES member and have something important to say about this paper then we urge you to join the AES today and make your voice heard. You can join online today by clicking here.