Filter bank design based on discriminative feature extraction

Author

Biem, Alain ; Katagiri, Shigeru

Author_Institution

ATR Human Inf. Process. Labs., Japan

fYear

1994

fDate

19-22 Apr 1994

Abstract

A filter bank model, which achieves minimum error, is investigated in this paper. A bank-of-filter feature extractor module is comprehensively optimized with the classifier´s parameters for minimization of the errors occurring at the back-end classifier. The method has been applied to readjusting Mel-scale and Bark-scale based filter banks for the Japanese vowel recognition task, the framework being provided by the minimum classification error (MCE)/generalised probabilistic descent method (GPD). The results show suggestive phenomena underlying the accuracy of the proposed approach

Keywords

band-pass filters; feature extraction; filtering theory; minimisation; probability; speech processing; speech recognition; Bark-scale filter banks; Japanese vowel recognition; Mel-scale filter banks; accuracy; back-end classifier; discriminative feature extraction; feature extractor module; filter bank design; filter bank model; generalised probabilistic descent method; minimum classification error; minimum error; Band pass filters; Bandwidth; Feature extraction; Filter bank; Filtering; Frequency conversion; Humans; Information processing; Speech processing; Speech recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1994. ICASSP-94., 1994 IEEE International Conference on

Conference_Location

Adelaide, SA

ISSN

1520-6149

Print_ISBN

0-7803-1775-0

Type

conf

DOI

10.1109/ICASSP.1994.389250

Filename

389250