Comments on Vocal Tract Length Normalization Equals Linear Transformation in Cepstral Space

Author

Afify, Mohamed ; Siohan, Olivier

Author_Institution

IBM T. J. Watson Res. Center, Yorktown Heights

Volume

15

Issue

5

fYear

2007

fDate

7/1/2007 12:00:00 AM

Firstpage

1731

Lastpage

1732

Abstract

The bilinear transformation (BT) is used for vocal tract length normalization (VTLN) in speech recogniton systems. We prove two properties of the bilinear mapping that motivated the band-diagonal transform proposed in M. Afify and O. Siohan, (ldquoConstrained maximum likelihood linear regression for speaker adaptation,rdquo in Proc. ICSLP, Beijing, China, Oct. 2000.) This is in contrast to what is stated in M. Pitz and H. Ney, (ldquoVocal tract length normalization equals linear transformation in cepstral space,rdquo IEEE Transactions on Speech and Audio Processing, vol. 13, no. 5, pp 930-944, September 2005) that the transform of Afify and Siohan was motivated by empirical observations.

Keywords

cepstral analysis; maximum likelihood estimation; regression analysis; speaker recognition; speech processing; audio processing; band-diagonal transform; bilinear transformation; cepstral space; linear transformation; maximum likelihood linear regression; speaker adaptation; speech processing; speech recogniton system; vocal tract length normalization; Adaptation model; Cepstral analysis; Equations; Frequency; Linear regression; Maximum likelihood linear regression; Natural languages; Speech processing; Speech recognition; Transforms; Maximum-likelihood linear regression; speaker adaptation; speech recognition; vocal tract length normalization;

fLanguage

English

Journal_Title

Audio, Speech, and Language Processing, IEEE Transactions on

Publisher

ieee

ISSN

1558-7916

Type

jour

DOI

10.1109/TASL.2007.896653

Filename

4244505