Date of Award

Summer 2001

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Electrical & Computer Engineering

Program/Concentration

Electrical Engineering

Committee Director

Stephen A. Zahorian

Committee Member

Lee A. Belfore II

Committee Member

Amin N. Dharamsi

Call Number for Print

Special Collections LD4331.E55 V46 2001

Abstract

Spectral feature computations continue to be a very difficult problem for accurate machine recognition of speech. In this work, which focuses on vowels, a new spectral peak envelope method for vowel classification is developed, based on a missing frequency components model of speech recognition. According to the missing frequency components model, vowel recognition depends only on the spectral (harmonic) peaks. Smoothing and interpolation of the spectra, performed in the standard cepstral analysis method commonly used in automatic speech recognition, actually loses valuable information and results in reduced recognition accuracy. The new method for feature extraction presented in this thesis is based on minimum mean square error curve fitting of cosine-like basis vectors to all peaks in the speech spectrum. A mathematical model for smoothly tracking spectral envelopes using only spectral peak information and ignoring other parts of the spectrum is presented. A software algorithm in Matlab for the model was developed and tested for various speaker types using a neural network classifier. Vowel classification experiments were conducted based on the features derived from the spectral peaks. The classification rates of the peak method under various signal to noise ratios was also studied. The basic conclusion is that the new features perform about the same as cepstral (also referred to as DCTCs, or Discrete Cosine Transform Coefficients) features for clean speech, but have advantages when the signal is degraded by noise.

Rights

In Copyright. URI: http://rightsstatements.org/vocab/InC/1.0/ This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).

DOI

10.25777/chns-hw82

Download

Share

COinS