A wavelet filter optimization algorithm for speech recognition

Author

Kaisheng, Yao ; Zhigang, Cao

Author_Institution

Dept. of Electron. Eng., Tsinghua Univ., Beijing, China

fYear

1998

fDate

22-24 Oct 1998

Abstract

A wavelet filter optimization algorithm is proposed. The proposed algorithm is based on wavelet matrix parameterization, through which wavelet filters can be factorized to a smaller set of parameters. Based on the cross validation cost function defined in this paper, Monte Carlo simulation is used to search the optimal wavelet filters for speech recognition. Using this wavelet filter optimization algorithm, we can obtain optimal wavelet filters for an isolated word recognition system. Also in this paper, mel wavelet cepstral coefficients, the speech feature generated by the optimized symmetrical shift-invariant wavelet, is described. Simulation results show that the continuous density HMM based isolated word recognition system can be optimized to achieve a recognition rate of 100% for 10 digits

Keywords

Monte Carlo methods; circuit optimisation; digital simulation; filtering theory; hidden Markov models; speech recognition; wavelet transforms; Monte Carlo simulation; continuous density HMM; cross validation cost function; isolated word recognition system; mel wavelet cepstral coefficients; optimal wavelet filters; optimized symmetrical shift-invariant wavelet; recognition rate; simulation results; speech feature; speech recognition; wavelet filter optimization algorithm; wavelet filters; wavelet matrix parameterization; Cepstral analysis; Discrete wavelet transforms; Filters; Hidden Markov models; Signal processing; Signal processing algorithms; Signal resolution; Speech recognition; Symmetric matrices; Wavelet domain;

fLanguage

English

Publisher

ieee

Conference_Titel

Communication Technology Proceedings, 1998. ICCT '98. 1998 International Conference on

Conference_Location

Beijing

Print_ISBN

7-80090-827-5

Type

conf

DOI

10.1109/ICCT.1998.741008

Filename

741008