Perceptual MVDR-based unsupervised built-in speaker normalization for Kazakh speech recognition

Author

Yessenbayev, Zhandos ; Yapanel, Umit

Author_Institution

Nazarbayev Univ. Res. & Innovation Syst., Astana, Kazakhstan

fYear

2014

fDate

15-17 Oct. 2014

Firstpage

1

Lastpage

5

Abstract

In this work we present a novel approach to unsupervised speaker normalization on top of the Perceptual MVDR-based Built-in Speaker Normalization technique. We showed that the proposed method can be efficient for the task of phonetic recognition on TIMIT and then applied it to Kazakh speech recognition. From the experiments, we see that this method is able to improve the relative performance of ASR systems up to 20% The analysis of the optimal warp factor selection by the algorithm revealed a nice gender separation ability which may be used for gender/speaker classification tasks.

Keywords

natural language processing; speech recognition; ASR systems; Kazakh speech recognition; TIMIT; gender classification tasks; gender separation ability; optimal warp factor selection; perceptual MVDR-based unsupervised built-in speaker normalization; phonetic recognition; speaker classification tasks; Acoustics; Algorithm design and analysis; Feature extraction; Hidden Markov models; Speech; Speech recognition; Training; Kazakh speech recognition; Unsupervised speaker normalization; phone recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Application of Information and Communication Technologies (AICT), 2014 IEEE 8th International Conference on

Conference_Location

Astana

Print_ISBN

978-1-4799-4120-9

Type

conf

DOI

10.1109/ICAICT.2014.7035914

Filename

7035914