Improving the filter bank of a classic speech feature extraction algorithm

Author

Skowronski, Mark D. ; Harris, John G.

Author_Institution

Computational Neuro-Eng. Lab, Florida Univ., Gainesville, FL, USA

Volume

4

fYear

2003

fDate

25-28 May 2003

Abstract

The most popular speech feature extractor used in automatic speech recognition (ASR) systems today is the mel frequency cepstral coefficient (MFCC) algorithm. Introduced in 1980, the filter bank-based algorithm eventually replaced linear prediction cepstral coefficients (LPCC) as the premier front end, primarily because of MFCC´s superior robustness to additive noise. However, MFCC does not approximate the critical bandwidth of the human auditory system. We propose a novel scheme for decoupling filter bandwidth from other filter bank parameters, and we demonstrate improved noise robustness over three versions of MFCC through HMM-based experiments with the English digits in various noise environments.

Keywords

cepstral analysis; feature extraction; filtering theory; hidden Markov models; noise; speech recognition; HMM-based experiments; MFCC algorithm; automatic speech recognition systems; filter bandwidth decoupling; filter bank-based algorithm; human auditory system; mel frequency cepstral coefficient algorithm; noise environments; noise robustness; speech feature extraction algorithm; Additive noise; Automatic speech recognition; Bandwidth; Cepstral analysis; Feature extraction; Filter bank; Mel frequency cepstral coefficient; Noise robustness; Nonlinear filters; Working environment noise;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 2003. ISCAS '03. Proceedings of the 2003 International Symposium on

Print_ISBN

0-7803-7761-3

Type

conf

DOI

10.1109/ISCAS.2003.1205828

Filename

1205828