Mel-frequency cepstrum coefficient analysis of infant cry with hypothyroidism

Abstract

This paper presents a new method for the analysis of newborn infant cry to detect hypothyroidism. Hypothyroidism is a condition caused by insufficient production of thyroid hormone by the thyroid gland. The proposed technique is robust as it automatically extracts and analyzes features of infant cry signal. In pre-processing of the signals, separation of voice or unvoiced sections is implemented with automatic segmentation that integrates zero rate crossing and short time energy methods. For feature extraction, the Mel frequency cepstrum coefficient analysis is used to extract main features from the infant cry signal. Then, similarities between the two types of signals and significant information are studied using the F-Ratio analysis. Results show that the F-Ratio analysis was able to discriminate between essential and non-essential features for classification.