Speaker Accent Classification System using Fuzzy Canonical Correlation-Based Gaussian Classifier

Author

Ullah, Sameeh ; Karray, Fakhri

Author_Institution

Dept. of Electr. & Comput. Eng., Univ. of Waterloo, Waterloo, ON, Canada

fYear

2007

fDate

24-27 Nov. 2007

Firstpage

792

Lastpage

795

Abstract

A speaker´s accent is the most important factor affecting the performance of automatic speech recognition (ASR) systems because accents vary widely, even within the same country or community. The reason may be attributed to the fuzziness between the boundaries of phoneme classes, a result of differences in a speaker´s vocal tract and accent. In this paper, a new method is proposed that is based on the fuzzy canonical correlation-based Gaussian classifier. In the proposed method the membership values of the clusters are based not only on minimizing the distance between the cluster centroids, but also maximizing the out-of-class variations.

Keywords

Gaussian processes; fuzzy set theory; speaker recognition; Gaussian classifier; automatic speech recognition systems; cluster centroids; fuzzy canonical correlation; out-of-class variations; speaker accent classification system; Anatomy; Automatic speech recognition; Clustering algorithms; Degradation; Fuzzy systems; Hidden Markov models; Neural networks; Signal processing; Support vector machine classification; Support vector machines; Fuzzy canonical correlation; Gaussian classifier; speaker´s accent; vocal tract;

fLanguage

English

Publisher

ieee

Conference_Titel

Signal Processing and Communications, 2007. ICSPC 2007. IEEE International Conference on

Conference_Location

Dubai

Print_ISBN

978-1-4244-1235-8

Electronic_ISBN

978-1-4244-1236-5

Type

conf

DOI

10.1109/ICSPC.2007.4728438

Filename

4728438