Improved voice activity detection for speech recognition system

Siew Wen Chin, Kah Phooi Seng, Li Minn Ang, King Hann Lim

Research output: Book chapter/Published conference paperConference paper

6 Citations (Scopus)

Abstract

An improved voice activity detection (VAD) based on the radial basis function neural network (RBF NN) and continuous wavelet transform (CWT) for speech recognition system is presented in the paper. The input speech signal is analyzed in the form of fixed size window by using Mel-frequency cepstral coefficients (MFCC). Within the windowed signal, the proposed RBF-CWT VAD algorithm detects the speech/ non-speech signal using the RBF NN. Once the interchange of speech to non-speech or vice versa occurred, the energy changes of the CWT coefficients are calculated to localize the final coordination of the starting/ending speech points. Instead of classifying the speech signal using the MFCC at the frame-level which easily capture lots of undesired noise encountered by the conventional VAD with the binary classifier, the proposed RBF NN with the aid of CWT analyzes the transformation of the MFCC at the window-level that offers a better compensation to the noisy signal. The simulation results shows an improvement on the precision of the speech detection and the overall ASR rate particularly under the noisy circumstances compared to the conventional VAD with the zero-crossing rate, short-term signal energy and binary classifier.

Original languageEnglish
Title of host publicationICS 2010 - International Computer Symposium
Pages518-523
Number of pages6
DOIs
Publication statusPublished - 01 Dec 2010
Event2010 International Computer Symposium, ICS 2010 - Tainan, Taiwan, Province of China
Duration: 16 Dec 201018 Dec 2010

Conference

Conference2010 International Computer Symposium, ICS 2010
CountryTaiwan, Province of China
CityTainan
Period16/12/1018/12/10

Fingerprint

Speech recognition
Wavelet transforms
Neural networks
Classifiers
Interchanges

Cite this

Chin, S. W., Seng, K. P., Ang, L. M., & Lim, K. H. (2010). Improved voice activity detection for speech recognition system. In ICS 2010 - International Computer Symposium (pp. 518-523). [5685456] https://doi.org/10.1109/COMPSYM.2010.5685456
Chin, Siew Wen ; Seng, Kah Phooi ; Ang, Li Minn ; Lim, King Hann. / Improved voice activity detection for speech recognition system. ICS 2010 - International Computer Symposium. 2010. pp. 518-523
@inproceedings{111c586f44484e1bbf901dd3c31ca99a,
title = "Improved voice activity detection for speech recognition system",
abstract = "An improved voice activity detection (VAD) based on the radial basis function neural network (RBF NN) and continuous wavelet transform (CWT) for speech recognition system is presented in the paper. The input speech signal is analyzed in the form of fixed size window by using Mel-frequency cepstral coefficients (MFCC). Within the windowed signal, the proposed RBF-CWT VAD algorithm detects the speech/ non-speech signal using the RBF NN. Once the interchange of speech to non-speech or vice versa occurred, the energy changes of the CWT coefficients are calculated to localize the final coordination of the starting/ending speech points. Instead of classifying the speech signal using the MFCC at the frame-level which easily capture lots of undesired noise encountered by the conventional VAD with the binary classifier, the proposed RBF NN with the aid of CWT analyzes the transformation of the MFCC at the window-level that offers a better compensation to the noisy signal. The simulation results shows an improvement on the precision of the speech detection and the overall ASR rate particularly under the noisy circumstances compared to the conventional VAD with the zero-crossing rate, short-term signal energy and binary classifier.",
keywords = "Continuous wavelet transform, Mel frequency cepstral coefficient, Radial basis function, Voice activity detection",
author = "Chin, {Siew Wen} and Seng, {Kah Phooi} and Ang, {Li Minn} and Lim, {King Hann}",
year = "2010",
month = "12",
day = "1",
doi = "10.1109/COMPSYM.2010.5685456",
language = "English",
isbn = "9781424476404",
pages = "518--523",
booktitle = "ICS 2010 - International Computer Symposium",

}

Chin, SW, Seng, KP, Ang, LM & Lim, KH 2010, Improved voice activity detection for speech recognition system. in ICS 2010 - International Computer Symposium., 5685456, pp. 518-523, 2010 International Computer Symposium, ICS 2010, Tainan, Taiwan, Province of China, 16/12/10. https://doi.org/10.1109/COMPSYM.2010.5685456

Improved voice activity detection for speech recognition system. / Chin, Siew Wen; Seng, Kah Phooi; Ang, Li Minn; Lim, King Hann.

ICS 2010 - International Computer Symposium. 2010. p. 518-523 5685456.

Research output: Book chapter/Published conference paperConference paper

TY - GEN

T1 - Improved voice activity detection for speech recognition system

AU - Chin, Siew Wen

AU - Seng, Kah Phooi

AU - Ang, Li Minn

AU - Lim, King Hann

PY - 2010/12/1

Y1 - 2010/12/1

N2 - An improved voice activity detection (VAD) based on the radial basis function neural network (RBF NN) and continuous wavelet transform (CWT) for speech recognition system is presented in the paper. The input speech signal is analyzed in the form of fixed size window by using Mel-frequency cepstral coefficients (MFCC). Within the windowed signal, the proposed RBF-CWT VAD algorithm detects the speech/ non-speech signal using the RBF NN. Once the interchange of speech to non-speech or vice versa occurred, the energy changes of the CWT coefficients are calculated to localize the final coordination of the starting/ending speech points. Instead of classifying the speech signal using the MFCC at the frame-level which easily capture lots of undesired noise encountered by the conventional VAD with the binary classifier, the proposed RBF NN with the aid of CWT analyzes the transformation of the MFCC at the window-level that offers a better compensation to the noisy signal. The simulation results shows an improvement on the precision of the speech detection and the overall ASR rate particularly under the noisy circumstances compared to the conventional VAD with the zero-crossing rate, short-term signal energy and binary classifier.

AB - An improved voice activity detection (VAD) based on the radial basis function neural network (RBF NN) and continuous wavelet transform (CWT) for speech recognition system is presented in the paper. The input speech signal is analyzed in the form of fixed size window by using Mel-frequency cepstral coefficients (MFCC). Within the windowed signal, the proposed RBF-CWT VAD algorithm detects the speech/ non-speech signal using the RBF NN. Once the interchange of speech to non-speech or vice versa occurred, the energy changes of the CWT coefficients are calculated to localize the final coordination of the starting/ending speech points. Instead of classifying the speech signal using the MFCC at the frame-level which easily capture lots of undesired noise encountered by the conventional VAD with the binary classifier, the proposed RBF NN with the aid of CWT analyzes the transformation of the MFCC at the window-level that offers a better compensation to the noisy signal. The simulation results shows an improvement on the precision of the speech detection and the overall ASR rate particularly under the noisy circumstances compared to the conventional VAD with the zero-crossing rate, short-term signal energy and binary classifier.

KW - Continuous wavelet transform

KW - Mel frequency cepstral coefficient

KW - Radial basis function

KW - Voice activity detection

UR - http://www.scopus.com/inward/record.url?scp=79851483774&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79851483774&partnerID=8YFLogxK

U2 - 10.1109/COMPSYM.2010.5685456

DO - 10.1109/COMPSYM.2010.5685456

M3 - Conference paper

AN - SCOPUS:79851483774

SN - 9781424476404

SP - 518

EP - 523

BT - ICS 2010 - International Computer Symposium

ER -

Chin SW, Seng KP, Ang LM, Lim KH. Improved voice activity detection for speech recognition system. In ICS 2010 - International Computer Symposium. 2010. p. 518-523. 5685456 https://doi.org/10.1109/COMPSYM.2010.5685456