Bayesian feature enhancement using a mixture of unscented transformations for uncertainty decoding of noisy speech

Yusuke Shinohara; Masami Akamine

doi:10.1109/ICASSP.2009.4960647

Bayesian feature enhancement using a mixture of unscented transformations for uncertainty decoding of noisy speech

Yusuke Shinohara, Masami Akamine

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

11 Citations (Scopus)

Abstract

A new parameter estimation method for the Model-Based Feature Enhancement (MBFE) is presented. The conventional MBFE uses the vector Taylor series to calculate the parameters of non-linearly transformed distributions, though the linearization leads to a degraded performance. We use the unscented transformation to estimate the parameters, where a minimal number of samples propagated through the nonlinear transformation are used. By avoiding the linearization, the parameters are estimated more accurately. Experimental results on Aurora2 show that the proposed method reduces the word error rate by 8.48% relatively, while the computational cost is just modestly higher, compared with the conventional MBFE.

Original language	English
Title of host publication	2009 IEEE International Conference on Acoustics, Speech, and Signal Processing - Proceedings, ICASSP 2009
Pages	4569-4572
Number of pages	4
DOIs	https://doi.org/10.1109/ICASSP.2009.4960647
Publication status	Published - 2009
Externally published	Yes
Event	2009 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2009 - Taipei, Taiwan, Province of China Duration: 2009 Apr 19 → 2009 Apr 24

Publication series

Name	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
ISSN (Print)	1520-6149

Other

Other	2009 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2009
Country/Territory	Taiwan, Province of China
City	Taipei
Period	09/4/19 → 09/4/24

Keywords

Feature enhancement
Noisy speech recognition
Uncertainty decoding
Unscented transformation
Vector Taylor series

ASJC Scopus subject areas

Software
Signal Processing
Electrical and Electronic Engineering

Access to Document

10.1109/ICASSP.2009.4960647

Cite this

Shinohara, Y., & Akamine, M. (2009). Bayesian feature enhancement using a mixture of unscented transformations for uncertainty decoding of noisy speech. In 2009 IEEE International Conference on Acoustics, Speech, and Signal Processing - Proceedings, ICASSP 2009 (pp. 4569-4572). [4960647] (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). https://doi.org/10.1109/ICASSP.2009.4960647

Bayesian feature enhancement using a mixture of unscented transformations for uncertainty decoding of noisy speech. / Shinohara, Yusuke; Akamine, Masami.

2009 IEEE International Conference on Acoustics, Speech, and Signal Processing - Proceedings, ICASSP 2009. 2009. p. 4569-4572 4960647 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Shinohara, Y & Akamine, M 2009, Bayesian feature enhancement using a mixture of unscented transformations for uncertainty decoding of noisy speech. in 2009 IEEE International Conference on Acoustics, Speech, and Signal Processing - Proceedings, ICASSP 2009., 4960647, ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, pp. 4569-4572, 2009 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2009, Taipei, Taiwan, Province of China, 09/4/19. https://doi.org/10.1109/ICASSP.2009.4960647

Shinohara Y, Akamine M. Bayesian feature enhancement using a mixture of unscented transformations for uncertainty decoding of noisy speech. In 2009 IEEE International Conference on Acoustics, Speech, and Signal Processing - Proceedings, ICASSP 2009. 2009. p. 4569-4572. 4960647. (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). doi: 10.1109/ICASSP.2009.4960647

Shinohara, Yusuke ; Akamine, Masami. / Bayesian feature enhancement using a mixture of unscented transformations for uncertainty decoding of noisy speech. 2009 IEEE International Conference on Acoustics, Speech, and Signal Processing - Proceedings, ICASSP 2009. 2009. pp. 4569-4572 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

@inproceedings{498e0fb451c345bda4e2e3bba8e9406a,

title = "Bayesian feature enhancement using a mixture of unscented transformations for uncertainty decoding of noisy speech",

abstract = "A new parameter estimation method for the Model-Based Feature Enhancement (MBFE) is presented. The conventional MBFE uses the vector Taylor series to calculate the parameters of non-linearly transformed distributions, though the linearization leads to a degraded performance. We use the unscented transformation to estimate the parameters, where a minimal number of samples propagated through the nonlinear transformation are used. By avoiding the linearization, the parameters are estimated more accurately. Experimental results on Aurora2 show that the proposed method reduces the word error rate by 8.48% relatively, while the computational cost is just modestly higher, compared with the conventional MBFE.",

keywords = "Feature enhancement, Noisy speech recognition, Uncertainty decoding, Unscented transformation, Vector Taylor series",

author = "Yusuke Shinohara and Masami Akamine",

year = "2009",

doi = "10.1109/ICASSP.2009.4960647",

language = "English",

isbn = "9781424423545",

series = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

pages = "4569--4572",

booktitle = "2009 IEEE International Conference on Acoustics, Speech, and Signal Processing - Proceedings, ICASSP 2009",

note = "2009 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2009 ; Conference date: 19-04-2009 Through 24-04-2009",

}

TY - GEN

T1 - Bayesian feature enhancement using a mixture of unscented transformations for uncertainty decoding of noisy speech

AU - Shinohara, Yusuke

AU - Akamine, Masami

PY - 2009

Y1 - 2009

N2 - A new parameter estimation method for the Model-Based Feature Enhancement (MBFE) is presented. The conventional MBFE uses the vector Taylor series to calculate the parameters of non-linearly transformed distributions, though the linearization leads to a degraded performance. We use the unscented transformation to estimate the parameters, where a minimal number of samples propagated through the nonlinear transformation are used. By avoiding the linearization, the parameters are estimated more accurately. Experimental results on Aurora2 show that the proposed method reduces the word error rate by 8.48% relatively, while the computational cost is just modestly higher, compared with the conventional MBFE.

AB - A new parameter estimation method for the Model-Based Feature Enhancement (MBFE) is presented. The conventional MBFE uses the vector Taylor series to calculate the parameters of non-linearly transformed distributions, though the linearization leads to a degraded performance. We use the unscented transformation to estimate the parameters, where a minimal number of samples propagated through the nonlinear transformation are used. By avoiding the linearization, the parameters are estimated more accurately. Experimental results on Aurora2 show that the proposed method reduces the word error rate by 8.48% relatively, while the computational cost is just modestly higher, compared with the conventional MBFE.

KW - Feature enhancement

KW - Noisy speech recognition

KW - Uncertainty decoding

KW - Unscented transformation

KW - Vector Taylor series

UR - http://www.scopus.com/inward/record.url?scp=70349206345&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=70349206345&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2009.4960647

DO - 10.1109/ICASSP.2009.4960647

M3 - Conference contribution

AN - SCOPUS:70349206345

SN - 9781424423545

T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

SP - 4569

EP - 4572

BT - 2009 IEEE International Conference on Acoustics, Speech, and Signal Processing - Proceedings, ICASSP 2009

T2 - 2009 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2009

Y2 - 19 April 2009 through 24 April 2009

ER -

Bayesian feature enhancement using a mixture of unscented transformations for uncertainty decoding of noisy speech

Abstract

Publication series

Other

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this