Improving speaker verification in noisy environments using adaptive filtering and hybrid classification technique

Research output: Contribution to journalArticle

9 Citations (Scopus)

Abstract

This study describes two approaches of improving speaker verification in noisy environments. The first approach is implementation of a speaker verification classification technique base on hybrid Vector Quantization (VQ) and Hidden Markov Models (HMMs) in clean and noisy environments. The second approach is implementation of Adaptive Noise Cancelation (ANC) as pre-processing for noise removal. The motivation to implement hybrid classification technique is to improve the HMMs performance. It is shown that, by using me hybrid technique, an Equal Error Rate (EER) of 11.72% is achieved compared to HMM alone, which achieved 16.66% in clean environments. However, both techniques show degradation in noisy environments. In order to address these problems, an Adaptive Noise Cancellation (ANC) technique using adaptive filtering is implemented in the pre-processing stage due to its ability to separate overlapping speech frequency bands. Investigations using Least-Mean-Square (LMS), Normalized Least-Mean-Square (NLMS) and Recursive LeastSquares (RLS) adaptive filtering are conducted to find the best solution for the speaker verification system.

Original languageEnglish
Pages (from-to)107-115
Number of pages9
JournalInformation Technology Journal
Volume9
Issue number1
DOIs
Publication statusPublished - 2010

Fingerprint

Adaptive filtering
Hidden Markov models
Vector quantization
Processing
Frequency bands
Degradation

Keywords

  • Adaptive filtering least mean-square
  • Normalized least mean-square
  • Recursive least squares
  • Vector quantization hidden markov models

ASJC Scopus subject areas

  • Computer Science (miscellaneous)

Cite this

@article{8f6cb70e44ff4c119816504bf3bb4079,
title = "Improving speaker verification in noisy environments using adaptive filtering and hybrid classification technique",
abstract = "This study describes two approaches of improving speaker verification in noisy environments. The first approach is implementation of a speaker verification classification technique base on hybrid Vector Quantization (VQ) and Hidden Markov Models (HMMs) in clean and noisy environments. The second approach is implementation of Adaptive Noise Cancelation (ANC) as pre-processing for noise removal. The motivation to implement hybrid classification technique is to improve the HMMs performance. It is shown that, by using me hybrid technique, an Equal Error Rate (EER) of 11.72{\%} is achieved compared to HMM alone, which achieved 16.66{\%} in clean environments. However, both techniques show degradation in noisy environments. In order to address these problems, an Adaptive Noise Cancellation (ANC) technique using adaptive filtering is implemented in the pre-processing stage due to its ability to separate overlapping speech frequency bands. Investigations using Least-Mean-Square (LMS), Normalized Least-Mean-Square (NLMS) and Recursive LeastSquares (RLS) adaptive filtering are conducted to find the best solution for the speaker verification system.",
keywords = "Adaptive filtering least mean-square, Normalized least mean-square, Recursive least squares, Vector quantization hidden markov models",
author = "Ilyas, {M. Z.} and {Abdul Samad}, Salina and Aini Hussain and Ishak, {Khairul Anuar}",
year = "2010",
doi = "10.3923/itj.2010.107.115",
language = "English",
volume = "9",
pages = "107--115",
journal = "Information Technology Journal",
issn = "1812-5638",
publisher = "Asian Network for Scientific Information",
number = "1",

}

TY - JOUR

T1 - Improving speaker verification in noisy environments using adaptive filtering and hybrid classification technique

AU - Ilyas, M. Z.

AU - Abdul Samad, Salina

AU - Hussain, Aini

AU - Ishak, Khairul Anuar

PY - 2010

Y1 - 2010

N2 - This study describes two approaches of improving speaker verification in noisy environments. The first approach is implementation of a speaker verification classification technique base on hybrid Vector Quantization (VQ) and Hidden Markov Models (HMMs) in clean and noisy environments. The second approach is implementation of Adaptive Noise Cancelation (ANC) as pre-processing for noise removal. The motivation to implement hybrid classification technique is to improve the HMMs performance. It is shown that, by using me hybrid technique, an Equal Error Rate (EER) of 11.72% is achieved compared to HMM alone, which achieved 16.66% in clean environments. However, both techniques show degradation in noisy environments. In order to address these problems, an Adaptive Noise Cancellation (ANC) technique using adaptive filtering is implemented in the pre-processing stage due to its ability to separate overlapping speech frequency bands. Investigations using Least-Mean-Square (LMS), Normalized Least-Mean-Square (NLMS) and Recursive LeastSquares (RLS) adaptive filtering are conducted to find the best solution for the speaker verification system.

AB - This study describes two approaches of improving speaker verification in noisy environments. The first approach is implementation of a speaker verification classification technique base on hybrid Vector Quantization (VQ) and Hidden Markov Models (HMMs) in clean and noisy environments. The second approach is implementation of Adaptive Noise Cancelation (ANC) as pre-processing for noise removal. The motivation to implement hybrid classification technique is to improve the HMMs performance. It is shown that, by using me hybrid technique, an Equal Error Rate (EER) of 11.72% is achieved compared to HMM alone, which achieved 16.66% in clean environments. However, both techniques show degradation in noisy environments. In order to address these problems, an Adaptive Noise Cancellation (ANC) technique using adaptive filtering is implemented in the pre-processing stage due to its ability to separate overlapping speech frequency bands. Investigations using Least-Mean-Square (LMS), Normalized Least-Mean-Square (NLMS) and Recursive LeastSquares (RLS) adaptive filtering are conducted to find the best solution for the speaker verification system.

KW - Adaptive filtering least mean-square

KW - Normalized least mean-square

KW - Recursive least squares

KW - Vector quantization hidden markov models

UR - http://www.scopus.com/inward/record.url?scp=77649311367&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77649311367&partnerID=8YFLogxK

U2 - 10.3923/itj.2010.107.115

DO - 10.3923/itj.2010.107.115

M3 - Article

VL - 9

SP - 107

EP - 115

JO - Information Technology Journal

JF - Information Technology Journal

SN - 1812-5638

IS - 1

ER -