Acoustic Analysis of Pathologies

From Infancy to Young Adulthood

Amy Neustein, Hemant A. Patil, Amy Neustein, Hemant A. Patil

• Publisher: De Gruyter
• Year: 2020
• eBook ISBN:
  9781501513169
• Topic: Technik & Maschinenbau
• Subtopic: Informatik Allgemein

This book provides the reader with empirical findings on innovative signal processing approaches to detecting pathologies in infant cries, by comparing new technological approaches to standard ones. The contributors examine novel approaches to machine adaptation to dysarthric speech.

1 Understanding infant cry analysis for pathology classification

Anshu Chittora

Healthark Wellness Solutions LLP, Ahmedabad

Hemant A. Patil

Dhirubhai Ambani Institute of Information and Communication Technology(DA-IICT), Gandhinagar

Note: Anshu Chittora is now at Healthark Wellness Solutions LLP, Ahmedabad, Gujarat, India. The work was done while the author was at DA-IICT Gandhinagar, India, and it does not contain any Healthark Wellness Solutions LLP proprietary information.

Abstract

Infants are difficult to understand as they cannot communicate their requirements. This motivates us to decode their language in meaningful interpretations so that adults can understand the requirements of their children. In this chapter, the cry analysis techniques used so far are discussed and some experiments in this direction are reported. Spectrographic analysis is also shown for its good performance in some pathological cases of infant cries. Along with this, what makes the infant cry analysis task a difficult task is also elaborated. Some results on infant cry analysis and classification work are reported in the later sections of this chapter.

Keywords: Cry modes, spectrographic analysis, higher order spectral analysis (HOSA), bispectrum, higher order singular value decomposition (HOSVD),

1.1 Introduction

The word infant is derived from the Latin word “Infan” which means speechless. Since infants cannot communicate using a language used by adults, they use cry as their communication language. Any language has linguistic and paralinguistic content in it, but the infant crying has only paralinguistic content. Crying is generated from a set of complicated and sophisticated physiological activities that involves coordination among the brain, respiratory and motor control, and the vocal system. It is considered that the crying helps in development of infant’s physiology by increasing the pulmonary (lung) capacity [1].

In order to understand the actual reasons of the crying of an infant and to measure the well-being of an infant, development of a tool is needed which can support the parents to take necessary actions and ensure the healthy development of an infant. This requires analysis of the infant cries as it is the only tool used by the infants to convey their emotional and physical states. Infant cry analysis is not limited to pediatrics, it requires inputs of neurologists, engineering and linguistics also. Infant cry analysis can unfold the correlations among these fields.

A team of Scandinavian researchers started the research in infant cry analysis by applying signal processing methods such as spectrographic analysis and acoustic analysis of the cry signals. For the initial twenty years, spectrographic analysis remained the primary tool for cry analysis. Later on, researchers tried automatic and semiautomatic methods using computer-based algorithms. Using these methods, researchers tried to classify and analyse different cry types such as hunger, pain and pleasure cries and few attempts have also been made towards pathological cry analysis. Compared to other fields of the speech signal analysis, the domain of infant cry analysis is comparatively less explored and researched.

Apart from the signal processing point of view, cry has also been studied from the perception point of view. It has been noticed that the mother of the infant can recognize her infant’s cry. However, recent research shows that recognition of an infant’s cry by the caregiver is dependent on the time spent by the caregiver with the infant irrespective of the caregiver’s gender [2]. How parents perceive the cr...

Table of contents

• Title Page
• Copyright
• Contents
• 1 Understanding infant cry analysis for pathology classification
• 2 Unsupervised auditory filterbank learning for infant cry classification
• 3 Acoustic and prosodic analysis of vocalizations of 18-month-old toddlers with autism spectrum disorder
• 4 Computer-aided speech therapy for dysarthric speakers: Statistical acoustic modeling for automated verification of pronunciation accuracy
• 5 Communication improves when human or computer listeners adapt to dysarthria
• 6 Role of music on infant developments
• Speech Technology and Text Mining in Medicine and Health Care