An Automatic Approach to Extract Features from the Infant’s Cry Signals
6 Pages Posted: 6 Aug 2019
Date Written: August 4, 2019
Abstract
Responding to a new-born baby who communicates only through his/her cry is the most challenging issue in baby health care environment. Our Research work proposes an automatic algorithm to extract better features which can be further used to estimate accurately why a baby is crying by the analysis of its cry pattern. The signal processing techniques is applied to inspect the sound of these cries, by which it is possible to find the particular feature that carry the details about the circumstances that stimulated the cry and the coefficients is computed over LABVIEW. Features such as fundamental frequency (f0), spectral feature - Mel frequency cepstral coefficients (MFCCS) and Entropy based feature - Multi band spectral entropy signature (MBSES) were extracted. The extracted cry features are constructive and sensible, which will steer to acceptable classification and detection performance. Decoding baby cry with an Automatic cry detection (ACD) device is essential which will support the mother's to know and respond to their baby's needs, and in particular for the physician to treat the diseased infant in the early stage. This application can be used in the Obstetrics and Gynecology Clinique.
Keywords: Cry signal, Features, MBSES, MFCCs, ACD, Sound Artifacts, Spectral, Infant’s health care, LABVIEW
Suggested Citation: Suggested Citation