Performance analysis of wavelet transform for unspoken words

Abstract

This paper aims to identifY unspoken words using fa cial muscle activity without audible signal. A novel approach of wavelet analysis using diff erent filter coefficients and the performance analysis of each filter coefficients along with classifiers is introduced. The technique is successfu lly used to classifY 5 vowels, month name, weekdays and is not sensitive to the variation in sp eed and the sp eaking style of the diff erent subjects and is used as recognition variables to recognize the unspoken words. Th is is an eff icient technique that measures the relative muscle activity of the articulatory fa cial muscles and it is shown that fo rmer systems go through austere performance degradation when new words are acquainted. Us ing wavelet, word error rate and classification rate achieve a noteworthy success. This paper presents wavelet analysis of surfa ce electromyogram (SEMG) of fo ur fa cial muscles to decompose the signal and identifY the words and classifying these words by three wavelets. Us ing three filter coefficients of Daubechies, Coiflets and Sy mlets, the unspoken word has been recognized. Th is represents the proportional muscle activity of the fo ur muscles. These are classified using k-nearest neighbor, probabilistic neural network and support vector machine to identifY the sp eech. It is urged that such applied science may be used fo r the user to give simple voiceless commands when conditioned fo r the sp ecific user.