國立虎尾科技大學 |

Language: English

Back

Linear transformations of features f...

Hwang, Andrew I.

Linear transformations of features for automatic speech recognition .

Record Type:	Language materials, printed : Monograph/item
Title/Author:	Linear transformations of features for automatic speech recognition ./
Author:	Hwang, Andrew I.
Description:	82 p.
Notes:	Source: Masters Abstracts International, Volume: 50-02, page: 1195.
Contained By:	Masters Abstracts International50-02.
Subject:	Engineering, Electronics and Electrical. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=1499661
ISBN:	9781124880990

Linear transformations of features for automatic speech recognition .
Hwang, Andrew I.

Linear transformations of features for automatic speech recognition . - 82 p.

Source: Masters Abstracts International, Volume: 50-02, page: 1195.

Thesis (M.S.)--State University of New York at Binghamton, 2011.

Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), Heteroschedastic Linear Discriminant Analysis (HLDA), and Independent Component Analysis (ICA) are the techniques used in this work to reduce features and improve data analysis for classification. They are used in conjunction with the Non-Linear Discriminant Analysis (NLDA), where a Neural Network (NN) reduces dimensions, and a linear transform is applied to the data extracted from the bottleneck layer. This combination improves phonetic classification performance compared to the individual use of NLDA or the particular linear transforms analyzed in this work. For phonetic recognition using a Hidden Markov Model (HMM), the PCA, LDA and HLDA are used in combination with the NLDA and compared using different numbers of HMM mixture components. The NLDA/PCA was found to be superior to NLDA with any other linear transform in terms of automatic speech recognition accuracy.

ISBN: 9781124880990Subjects--Topical Terms:

845382
Engineering, Electronics and Electrical.

Linear transformations of features for automatic speech recognition .
LDR:01909nam 2200289 4500 001 713048
005 20121003100356.5
008 121101s2011 ||||||||||||||||| ||eng d
020 $a 9781124880990
035 $a (UMI)AAI1499661
035 $a AAI1499661
040 $a UMI $c UMI
100 1 $a Hwang, Andrew I. $3 845659
245 1 0 $a Linear transformations of features for automatic speech recognition .
300 $a 82 p.
500 $a Source: Masters Abstracts International, Volume: 50-02, page: 1195.
500 $a Adviser: Stephen A. Zahorian.
502 $a Thesis (M.S.)--State University of New York at Binghamton, 2011.
520 $a Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), Heteroschedastic Linear Discriminant Analysis (HLDA), and Independent Component Analysis (ICA) are the techniques used in this work to reduce features and improve data analysis for classification. They are used in conjunction with the Non-Linear Discriminant Analysis (NLDA), where a Neural Network (NN) reduces dimensions, and a linear transform is applied to the data extracted from the bottleneck layer. This combination improves phonetic classification performance compared to the individual use of NLDA or the particular linear transforms analyzed in this work. For phonetic recognition using a Hidden Markov Model (HMM), the PCA, LDA and HLDA are used in combination with the NLDA and compared using different numbers of HMM mixture components. The NLDA/PCA was found to be superior to NLDA with any other linear transform in terms of automatic speech recognition accuracy.
590 $a School code: 0792.
650 4 $a Engineering, Electronics and Electrical. $3 845382
690 $a 0544
710 2 $a State University of New York at Binghamton. $b Electrical Engineering. $3 845660
773 0 $t Masters Abstracts International $g 50-02.
790 1 0 $a Zahorian, Stephen A., $e advisor
790 1 0 $a Fowler, Mark L. $e committee member
790 1 0 $a Li, Xiaohua $e committee member
790 $a 0792
791 $a M.S.
792 $a 2011
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=1499661

based on 0 review(s)

Multimedia

Reviews

Add a review and share your thoughts with other readers

Export

pickup library