Date of Award

Fall 1975

Document Type

Thesis - Restricted

Degree Name

Master of Science (MS)

Department

Electrical and Computer Engineering

First Advisor

Niederjohn, Russell J.

Second Advisor

Wu, S. H.

Third Advisor

Heinen, James A.

Abstract

A simulation analysis is described which Models the ten English vowels and determines features tor each of tour zero-crossing analysis techniques. In addition, a study is made of the applicability of these techniques to frequency discrimination among sine waves of various frequencies. The comparative analyses indicate the ability of the zero-crossing analysis techniques to discriminate the vowels and sine waves in both a noise-tree and Gaussian white noise environment. One of the four techniques, the h-method, appears to have the most desirable signal discrimination characteristics. The highest correct recognition rate was obtained for h-method feature sets, for realistic S/N ratios in both the sine wave and vowel analyses. It can sustain approximately 5 to 10 dB more noise than the other three techniques in the discrimination of the vowels, based upon a 90% correct recognition rate.

Share

COinS

Restricted Access Item

Having trouble?