Research on Speaker Recognition Based on Multifractal Spectrum Feature

Author

Zhou, Yuhuan ; Wang, Jinming ; Zhang, Xiongwei

Author_Institution

PLA Univ. of Sci. & Technol., Nanjing, China

Volume

1

fYear

2010

Firstpage

463

Lastpage

466

Abstract

In this paper, a new nonlinear feature extraction method based on the WTMM (wavelet transform modulus-maxima method) is proposed, which can greatly facilitate the extraction of the multifractal spectrum feature (MSF) from speech signals. The MSF combined with traditional linear features can obviously improve the performance of speaker recognition system. Experiment results show that 6-dimensional MSF combined with LPC make recognition accuracy increase 6.4 percentage points, and 6-dimensional MSF combined with MFCC, LPC make recognition accuracy increase 1.6 percentage points and reach 98.8% in short speech (2 seconds) speaker recognition.

Keywords

feature extraction; linear predictive coding; speaker recognition; spectral analysis; speech processing; wavelet transforms; multifractal spectrum feature; nonlinear feature extraction method; speaker recognition system; speech signals; wavelet transform modulus-maxima method; Data mining; Feature extraction; Fractals; Geometry; Linear predictive coding; Mel frequency cepstral coefficient; Speaker recognition; Speech analysis; Speech recognition; Wavelet transforms; LPC; MFCC; multifractal spectrum feature; speaker recognition; wavelet transform modulus-maxima method;

fLanguage

English

Publisher

ieee

Conference_Titel

Computer Modeling and Simulation, 2010. ICCMS '10. Second International Conference on

Conference_Location

Sanya, Hainan

Print_ISBN

978-1-4244-5642-0

Electronic_ISBN

978-1-4244-5643-7

Type

conf

DOI

10.1109/ICCMS.2010.66

Filename

5421347