Date of Award

Summer 1987

Document Type

Thesis - Restricted

Degree Name

Master of Science (MS)

Department

Electrical and Computer Engineering

First Advisor

Niederjohn, Russell J.

Second Advisor

Josse, Fabien

Third Advisor

Doerr, Thomas A.

Abstract

In this thesis, objective and subjective evaluations of the spectral subtraction method for enhancing speech corrupted by additive white noise are carried out by: 1) error calculations, 2) spectrograms, and 3) subjective listening tests. The experimental parameters consist of different signal-to-noise ratios (-18 to 12dB) and different parameters of the spectral subtraction method. Thirty utterances, which consisted of the combination of five sentences and six people (three male and three female), were used for the experiments. Also, the spectral subtraction algorithm is considered as a preprocessor for speech feature extraction algorithms. Both pitch extraction algorithms and formant extraction algorithms were tested.

Share

COinS

Restricted Access Item

Having trouble?