Optimizing feature extraction for speech recognition

Chulhee Lee, Donghoon Hyun, Euisun Choi, Jinwook Go, Chungyong Lee

Research output: Contribution to journalArticle

35 Citations (Scopus)

Abstract

In this paper, we propose a method to minimize the loss of information during the feature extraction stage in speech recognition by optimizing the parameters of the mel-cepstrum transformation, a transform which is widely used in speech recognition. Typically, the mel-cepstrum is obtained by critical band filters whose characteristics play an important role in converting a speech signal into a sequence of vectors. First, we analyze the performance of the mel-cepstrum by changing the parameters of the filters such as shape, center frequency, and bandwidth. Then we propose an algorithm to optimize the parameters of the filters using the simplex method. Experiments with Korean digit words show that the recognition rate improved by about 4-7%.

Original languageEnglish
Pages (from-to)80-87
Number of pages8
JournalIEEE Transactions on Speech and Audio Processing
Volume11
Issue number1
DOIs
Publication statusPublished - 2003 Jan 1

Fingerprint

speech recognition
Speech recognition
pattern recognition
Feature extraction
filters
simplex method
digits
Bandwidth
bandwidth
Experiments

All Science Journal Classification (ASJC) codes

  • Software
  • Acoustics and Ultrasonics
  • Computer Vision and Pattern Recognition
  • Electrical and Electronic Engineering

Cite this

Lee, Chulhee ; Hyun, Donghoon ; Choi, Euisun ; Go, Jinwook ; Lee, Chungyong. / Optimizing feature extraction for speech recognition. In: IEEE Transactions on Speech and Audio Processing. 2003 ; Vol. 11, No. 1. pp. 80-87.
@article{cd2835031e8646baa8703b9693b6e276,
title = "Optimizing feature extraction for speech recognition",
abstract = "In this paper, we propose a method to minimize the loss of information during the feature extraction stage in speech recognition by optimizing the parameters of the mel-cepstrum transformation, a transform which is widely used in speech recognition. Typically, the mel-cepstrum is obtained by critical band filters whose characteristics play an important role in converting a speech signal into a sequence of vectors. First, we analyze the performance of the mel-cepstrum by changing the parameters of the filters such as shape, center frequency, and bandwidth. Then we propose an algorithm to optimize the parameters of the filters using the simplex method. Experiments with Korean digit words show that the recognition rate improved by about 4-7{\%}.",
author = "Chulhee Lee and Donghoon Hyun and Euisun Choi and Jinwook Go and Chungyong Lee",
year = "2003",
month = "1",
day = "1",
doi = "10.1109/TSA.2002.805644",
language = "English",
volume = "11",
pages = "80--87",
journal = "IEEE Transactions on Speech and Audio Processing",
issn = "1063-6676",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "1",

}

Optimizing feature extraction for speech recognition. / Lee, Chulhee; Hyun, Donghoon; Choi, Euisun; Go, Jinwook; Lee, Chungyong.

In: IEEE Transactions on Speech and Audio Processing, Vol. 11, No. 1, 01.01.2003, p. 80-87.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Optimizing feature extraction for speech recognition

AU - Lee, Chulhee

AU - Hyun, Donghoon

AU - Choi, Euisun

AU - Go, Jinwook

AU - Lee, Chungyong

PY - 2003/1/1

Y1 - 2003/1/1

N2 - In this paper, we propose a method to minimize the loss of information during the feature extraction stage in speech recognition by optimizing the parameters of the mel-cepstrum transformation, a transform which is widely used in speech recognition. Typically, the mel-cepstrum is obtained by critical band filters whose characteristics play an important role in converting a speech signal into a sequence of vectors. First, we analyze the performance of the mel-cepstrum by changing the parameters of the filters such as shape, center frequency, and bandwidth. Then we propose an algorithm to optimize the parameters of the filters using the simplex method. Experiments with Korean digit words show that the recognition rate improved by about 4-7%.

AB - In this paper, we propose a method to minimize the loss of information during the feature extraction stage in speech recognition by optimizing the parameters of the mel-cepstrum transformation, a transform which is widely used in speech recognition. Typically, the mel-cepstrum is obtained by critical band filters whose characteristics play an important role in converting a speech signal into a sequence of vectors. First, we analyze the performance of the mel-cepstrum by changing the parameters of the filters such as shape, center frequency, and bandwidth. Then we propose an algorithm to optimize the parameters of the filters using the simplex method. Experiments with Korean digit words show that the recognition rate improved by about 4-7%.

UR - http://www.scopus.com/inward/record.url?scp=0037228174&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0037228174&partnerID=8YFLogxK

U2 - 10.1109/TSA.2002.805644

DO - 10.1109/TSA.2002.805644

M3 - Article

AN - SCOPUS:0037228174

VL - 11

SP - 80

EP - 87

JO - IEEE Transactions on Speech and Audio Processing

JF - IEEE Transactions on Speech and Audio Processing

SN - 1063-6676

IS - 1

ER -