A Time-Varying Adaptive IIR Filter for Robust Text-Independent Speaker Verification

Santi NURATCH; Panuthat BOONPRAMUK; Chai WUTIWIWATCHAI

doi:10.1587/transinf.E96.D.699

A Time-Varying Adaptive IIR Filter for Robust Text-Independent Speaker Verification

Santi NURATCH, Panuthat BOONPRAMUK, Chai WUTIWIWATCHAI

Full Text Views

0

Share
Cite this

Summary :

This paper presents a new technique to smooth speech feature vectors for text-independent speaker verification using an adaptive band-pass IIR filer. The filter is designed by considering the probability density of modulation-frequency components of an M-dimensional feature vector. Each dimension of the feature vector is processed and filtered separately. Initial filter parameters, low-cut-off and high-cut-off frequencies, are first determined by the global mean of the probability densities computed from all feature vectors of a given speech utterance. Then, the cut-off frequencies are adapted over time, i.e. every frame vector, in both low-frequency and high-frequency bands based also on the global mean and the standard deviation of feature vectors. The filtered feature vectors are used in a SVM-GMM Supervector speaker verification system. The NIST Speaker Recognition Evaluation 2006 (SRE06) core-test is used in evaluation. Experimental results show that the proposed technique clearly outperforms a baseline system using a conventional RelAtive SpecTrA (RASTA) filter.

Publication: IEICE TRANSACTIONS on Information Vol.E96-D No.3 pp.699-707

Publication Date: 2013/03/01

Publicized

Online ISSN: 1745-1361

DOI: 10.1587/transinf.E96.D.699

Type of Manuscript: PAPER

Category: Speech and Hearing

Cite this

Copy

Santi NURATCH, Panuthat BOONPRAMUK, Chai WUTIWIWATCHAI, "A Time-Varying Adaptive IIR Filter for Robust Text-Independent Speaker Verification" in IEICE TRANSACTIONS on Information, vol. E96-D, no. 3, pp. 699-707, March 2013, doi: 10.1587/transinf.E96.D.699.
Abstract: This paper presents a new technique to smooth speech feature vectors for text-independent speaker verification using an adaptive band-pass IIR filer. The filter is designed by considering the probability density of modulation-frequency components of an M-dimensional feature vector. Each dimension of the feature vector is processed and filtered separately. Initial filter parameters, low-cut-off and high-cut-off frequencies, are first determined by the global mean of the probability densities computed from all feature vectors of a given speech utterance. Then, the cut-off frequencies are adapted over time, i.e. every frame vector, in both low-frequency and high-frequency bands based also on the global mean and the standard deviation of feature vectors. The filtered feature vectors are used in a SVM-GMM Supervector speaker verification system. The NIST Speaker Recognition Evaluation 2006 (SRE06) core-test is used in evaluation. Experimental results show that the proposed technique clearly outperforms a baseline system using a conventional RelAtive SpecTrA (RASTA) filter.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.E96.D.699/_p

Copy

@ARTICLE{e96-d_3_699,
author={Santi NURATCH, Panuthat BOONPRAMUK, Chai WUTIWIWATCHAI, },
journal={IEICE TRANSACTIONS on Information},
title={A Time-Varying Adaptive IIR Filter for Robust Text-Independent Speaker Verification},
year={2013},
volume={E96-D},
number={3},
pages={699-707},
abstract={This paper presents a new technique to smooth speech feature vectors for text-independent speaker verification using an adaptive band-pass IIR filer. The filter is designed by considering the probability density of modulation-frequency components of an M-dimensional feature vector. Each dimension of the feature vector is processed and filtered separately. Initial filter parameters, low-cut-off and high-cut-off frequencies, are first determined by the global mean of the probability densities computed from all feature vectors of a given speech utterance. Then, the cut-off frequencies are adapted over time, i.e. every frame vector, in both low-frequency and high-frequency bands based also on the global mean and the standard deviation of feature vectors. The filtered feature vectors are used in a SVM-GMM Supervector speaker verification system. The NIST Speaker Recognition Evaluation 2006 (SRE06) core-test is used in evaluation. Experimental results show that the proposed technique clearly outperforms a baseline system using a conventional RelAtive SpecTrA (RASTA) filter.},
keywords={},
doi={10.1587/transinf.E96.D.699},
ISSN={1745-1361},
month={March},}

Copy

TY - JOUR
TI - A Time-Varying Adaptive IIR Filter for Robust Text-Independent Speaker Verification
T2 - IEICE TRANSACTIONS on Information
SP - 699
EP - 707
AU - Santi NURATCH
AU - Panuthat BOONPRAMUK
AU - Chai WUTIWIWATCHAI
PY - 2013
DO - 10.1587/transinf.E96.D.699
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E96-D
IS - 3
JA - IEICE TRANSACTIONS on Information
Y1 - March 2013
AB - This paper presents a new technique to smooth speech feature vectors for text-independent speaker verification using an adaptive band-pass IIR filer. The filter is designed by considering the probability density of modulation-frequency components of an M-dimensional feature vector. Each dimension of the feature vector is processed and filtered separately. Initial filter parameters, low-cut-off and high-cut-off frequencies, are first determined by the global mean of the probability densities computed from all feature vectors of a given speech utterance. Then, the cut-off frequencies are adapted over time, i.e. every frame vector, in both low-frequency and high-frequency bands based also on the global mean and the standard deviation of feature vectors. The filtered feature vectors are used in a SVM-GMM Supervector speaker verification system. The NIST Speaker Recognition Evaluation 2006 (SRE06) core-test is used in evaluation. Experimental results show that the proposed technique clearly outperforms a baseline system using a conventional RelAtive SpecTrA (RASTA) filter.
ER -