Improved Malay vowel feature extraction method based on first and second formants

M. Y. Shahrul Azmi, Fadzilah Siraj, S. Yaacob, M. P. Paulraj, Mohd Zakree Ahmad Nazri

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

There are many speech recognition applications that use vowels phonemes. Among them are speech therapy systems that improve utterances of word pronunciation especially to children. There are also systems that teach hearing impaired person to speak properly by pronouncing words with a good degree of intelligibility. All of these systems require high degree of vowel recognition capability. This paper presents a new method of Malay vowel feature extraction based on formant and spectrum envelope called First Formant Bandwidth (F1BW). It is an effort to increase Malay vowel recognition capability by using a new speech database that consist of words uttered by Malaysian speakers from the three major races, Malay, Chinese and Indians. Based on single frame analysis, F1BW performs better than MFCC by more than 9% based on four classifiers of Levenberg-Marquart trained Neural Network, K-Nearest Neighbours, Multinomial Logistic Regression and Linear Discriminant Analysis.

Original languageEnglish
Title of host publicationProceedings - 2nd International Conference on Computational Intelligence, Modelling and Simulation, CIMSim 2010
Pages339-344
Number of pages6
DOIs
Publication statusPublished - 2010
Externally publishedYes
Event2nd International Conference on Computational Intelligence, Modelling and Simulation, CIMSim 2010 - Bali
Duration: 28 Sep 201030 Sep 2010

Other

Other2nd International Conference on Computational Intelligence, Modelling and Simulation, CIMSim 2010
CityBali
Period28/9/1030/9/10

Fingerprint

Feature Extraction
Feature extraction
Audition
Discriminant analysis
Speech recognition
Logistics
Classifiers
Speech Recognition
Logistic Regression
Discriminant Analysis
Neural networks
Bandwidth
Envelope
Therapy
Nearest Neighbor
Person
Classifier
Neural Networks
Speech
Children

Keywords

  • Malay vowel
  • Spectrum envelope
  • Speech recognition

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Applied Mathematics
  • Modelling and Simulation

Cite this

Shahrul Azmi, M. Y., Siraj, F., Yaacob, S., Paulraj, M. P., & Ahmad Nazri, M. Z. (2010). Improved Malay vowel feature extraction method based on first and second formants. In Proceedings - 2nd International Conference on Computational Intelligence, Modelling and Simulation, CIMSim 2010 (pp. 339-344). [5701868] https://doi.org/10.1109/CIMSiM.2010.59

Improved Malay vowel feature extraction method based on first and second formants. / Shahrul Azmi, M. Y.; Siraj, Fadzilah; Yaacob, S.; Paulraj, M. P.; Ahmad Nazri, Mohd Zakree.

Proceedings - 2nd International Conference on Computational Intelligence, Modelling and Simulation, CIMSim 2010. 2010. p. 339-344 5701868.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Shahrul Azmi, MY, Siraj, F, Yaacob, S, Paulraj, MP & Ahmad Nazri, MZ 2010, Improved Malay vowel feature extraction method based on first and second formants. in Proceedings - 2nd International Conference on Computational Intelligence, Modelling and Simulation, CIMSim 2010., 5701868, pp. 339-344, 2nd International Conference on Computational Intelligence, Modelling and Simulation, CIMSim 2010, Bali, 28/9/10. https://doi.org/10.1109/CIMSiM.2010.59
Shahrul Azmi MY, Siraj F, Yaacob S, Paulraj MP, Ahmad Nazri MZ. Improved Malay vowel feature extraction method based on first and second formants. In Proceedings - 2nd International Conference on Computational Intelligence, Modelling and Simulation, CIMSim 2010. 2010. p. 339-344. 5701868 https://doi.org/10.1109/CIMSiM.2010.59
Shahrul Azmi, M. Y. ; Siraj, Fadzilah ; Yaacob, S. ; Paulraj, M. P. ; Ahmad Nazri, Mohd Zakree. / Improved Malay vowel feature extraction method based on first and second formants. Proceedings - 2nd International Conference on Computational Intelligence, Modelling and Simulation, CIMSim 2010. 2010. pp. 339-344
@inproceedings{8b92bc4dc82944828a1548994b30aa73,
title = "Improved Malay vowel feature extraction method based on first and second formants",
abstract = "There are many speech recognition applications that use vowels phonemes. Among them are speech therapy systems that improve utterances of word pronunciation especially to children. There are also systems that teach hearing impaired person to speak properly by pronouncing words with a good degree of intelligibility. All of these systems require high degree of vowel recognition capability. This paper presents a new method of Malay vowel feature extraction based on formant and spectrum envelope called First Formant Bandwidth (F1BW). It is an effort to increase Malay vowel recognition capability by using a new speech database that consist of words uttered by Malaysian speakers from the three major races, Malay, Chinese and Indians. Based on single frame analysis, F1BW performs better than MFCC by more than 9{\%} based on four classifiers of Levenberg-Marquart trained Neural Network, K-Nearest Neighbours, Multinomial Logistic Regression and Linear Discriminant Analysis.",
keywords = "Malay vowel, Spectrum envelope, Speech recognition",
author = "{Shahrul Azmi}, {M. Y.} and Fadzilah Siraj and S. Yaacob and Paulraj, {M. P.} and {Ahmad Nazri}, {Mohd Zakree}",
year = "2010",
doi = "10.1109/CIMSiM.2010.59",
language = "English",
isbn = "9780769542621",
pages = "339--344",
booktitle = "Proceedings - 2nd International Conference on Computational Intelligence, Modelling and Simulation, CIMSim 2010",

}

TY - GEN

T1 - Improved Malay vowel feature extraction method based on first and second formants

AU - Shahrul Azmi, M. Y.

AU - Siraj, Fadzilah

AU - Yaacob, S.

AU - Paulraj, M. P.

AU - Ahmad Nazri, Mohd Zakree

PY - 2010

Y1 - 2010

N2 - There are many speech recognition applications that use vowels phonemes. Among them are speech therapy systems that improve utterances of word pronunciation especially to children. There are also systems that teach hearing impaired person to speak properly by pronouncing words with a good degree of intelligibility. All of these systems require high degree of vowel recognition capability. This paper presents a new method of Malay vowel feature extraction based on formant and spectrum envelope called First Formant Bandwidth (F1BW). It is an effort to increase Malay vowel recognition capability by using a new speech database that consist of words uttered by Malaysian speakers from the three major races, Malay, Chinese and Indians. Based on single frame analysis, F1BW performs better than MFCC by more than 9% based on four classifiers of Levenberg-Marquart trained Neural Network, K-Nearest Neighbours, Multinomial Logistic Regression and Linear Discriminant Analysis.

AB - There are many speech recognition applications that use vowels phonemes. Among them are speech therapy systems that improve utterances of word pronunciation especially to children. There are also systems that teach hearing impaired person to speak properly by pronouncing words with a good degree of intelligibility. All of these systems require high degree of vowel recognition capability. This paper presents a new method of Malay vowel feature extraction based on formant and spectrum envelope called First Formant Bandwidth (F1BW). It is an effort to increase Malay vowel recognition capability by using a new speech database that consist of words uttered by Malaysian speakers from the three major races, Malay, Chinese and Indians. Based on single frame analysis, F1BW performs better than MFCC by more than 9% based on four classifiers of Levenberg-Marquart trained Neural Network, K-Nearest Neighbours, Multinomial Logistic Regression and Linear Discriminant Analysis.

KW - Malay vowel

KW - Spectrum envelope

KW - Speech recognition

UR - http://www.scopus.com/inward/record.url?scp=79952089984&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79952089984&partnerID=8YFLogxK

U2 - 10.1109/CIMSiM.2010.59

DO - 10.1109/CIMSiM.2010.59

M3 - Conference contribution

SN - 9780769542621

SP - 339

EP - 344

BT - Proceedings - 2nd International Conference on Computational Intelligence, Modelling and Simulation, CIMSim 2010

ER -