An Approach for Computing Acoustic Coefficients of Dysarthric Speech In Kannada Language

Authors

  • Latha M, Dr M Shivakumar, Dr. Manjula.R

Abstract

The classification of speech sounds in kannada language depends on the set of phonemes which describes the  distinctive features with respect to articulatory,acoustic and phonetic properties. The utteranaces of any speech sounds in kannada langauge can be  identified through the estimation of  acoustic coefficients.The following bisyllabic words such as /??/, /??/, /??/, /??/ in Kannada language are assessed using MFCC Feature Extraction Method to derive the information of the speech signal at low frequency in the form of the cepstral coefficients of order 13 and are represented as acoustic feature vectors. These feature vectors are helpful in assessing phoneme significance of the uttered words which inturn represents the vocal tract resonances in the form of frequencies. The present work focuses on extracting the acoustic characteristics of Normal subjects and subjects with dysarthria using MFCC Feature Extraction Method. The results provides the detailed information about the pathological subjects  in more efficient way at lower frequency and contributes in the development of efficient  speech recognition systems.

Published

2020-11-01

Issue

Section

Articles