Date of Award

Fall 1976

Document Type

Thesis - Restricted

Degree Name

Master of Science (MS)

Department

Electrical and Computer Engineering

First Advisor

Niederjohn, Russell J.

Second Advisor

Wu, Sherman H.

Third Advisor

Horgan, James D.

Abstract

The development and testing of a speaker independent isolated spoken digit recognition system is described. The technique used involves contour analysis of both voicing and friction. The speech waveform is filtered and rectified to obtain the voicing and friction contours. An approximate frequency contour is obtained by utilizing a measure of the zero-crossing counts. By analysis of the voicing, friction, and zero-crossing contours a decision is made as to which digit was spoken. The sampling, storage, and analysis are all performed in real-time with a small minicomputer and a simple system of hardware circuits. The simple hardware circuits and small computer make this technique relatively inexpensive to implement as compared to other spoken digit recognition systems.

Share

COinS

Restricted Access Item

Having trouble?