Dereverberation with an Iterative Least-Squares Technique and Minimum Mean-Square Error Estimation for Automatic Speech Recognition

F. Mueller; A. Mertins

Dereverberation with an Iterative Least-Squares Technique and Minimum Mean-Square Error Estimation for Automatic Speech Recognition

Institute of Signal Processing

Abstract

This work is about dereverberation for automatic speech recognition. The use of a linear minimum mean-square error estimator for enhancing a recently proposed dereverberation method is investigated. The conducted phoneme recognition experiments show that the resynthesis step, which was done in the original work of the dereverberation method, can be omitted. Furthermore, it is shown that the recognition performance can be increased with the proposed estimator approach under certain reverberant conditions.

Original language	English
Title of host publication	Speech Communication; 10. ITG Symposium
Number of pages	4
Publisher	IEEE
Publication date	01.09.2012
Pages	1-4
Article number	6309625
ISBN (Electronic)	978-3-8007-3455-9
Publication status	Published - 01.09.2012
Event	10th ITG Symposium on Speech Communication - Braunschweig, Germany Duration: 26.09.2012 → 28.09.2012 Conference number: 116047

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

https://ieeexplore.ieee.org/document/6309625

Cite this

@inproceedings{ce975f6b83924974b219f7a30eb1404e,

title = "Dereverberation with an Iterative Least-Squares Technique and Minimum Mean-Square Error Estimation for Automatic Speech Recognition",

abstract = "This work is about dereverberation for automatic speech recognition. The use of a linear minimum mean-square error estimator for enhancing a recently proposed dereverberation method is investigated. The conducted phoneme recognition experiments show that the resynthesis step, which was done in the original work of the dereverberation method, can be omitted. Furthermore, it is shown that the recognition performance can be increased with the proposed estimator approach under certain reverberant conditions.",

author = "F. Mueller and A. Mertins",

year = "2012",

month = sep,

day = "1",

language = "English",

series = "IEEE Proceedings",

publisher = "IEEE",

pages = "1--4",

booktitle = "Speech Communication; 10. ITG Symposium",

note = "10th ITG Symposium on Speech Communication, ITGspeech 2012 ; Conference date: 26-09-2012 Through 28-09-2012",

}

Mueller, F & Mertins, A 2012, Dereverberation with an Iterative Least-Squares Technique and Minimum Mean-Square Error Estimation for Automatic Speech Recognition. in Speech Communication; 10. ITG Symposium., 6309625, IEEE Proceedings, IEEE, pp. 1-4, 10th ITG Symposium on Speech Communication, Braunschweig, Lower Saxony, Germany, 26.09.12. <https://ieeexplore.ieee.org/document/6309625>

TY - GEN

T1 - Dereverberation with an Iterative Least-Squares Technique and Minimum Mean-Square Error Estimation for Automatic Speech Recognition

AU - Mueller, F.

AU - Mertins, A.

N1 - Conference code: 116047

PY - 2012/9/1

Y1 - 2012/9/1

N2 - This work is about dereverberation for automatic speech recognition. The use of a linear minimum mean-square error estimator for enhancing a recently proposed dereverberation method is investigated. The conducted phoneme recognition experiments show that the resynthesis step, which was done in the original work of the dereverberation method, can be omitted. Furthermore, it is shown that the recognition performance can be increased with the proposed estimator approach under certain reverberant conditions.

AB - This work is about dereverberation for automatic speech recognition. The use of a linear minimum mean-square error estimator for enhancing a recently proposed dereverberation method is investigated. The conducted phoneme recognition experiments show that the resynthesis step, which was done in the original work of the dereverberation method, can be omitted. Furthermore, it is shown that the recognition performance can be increased with the proposed estimator approach under certain reverberant conditions.

UR - https://www.scopus.com/record/display.uri?eid=2-s2.0-84947967221&origin=inward&txGid=d0e39b7b3afa222fd6c14e8d6e77e90d

M3 - Conference contribution

T3 - IEEE Proceedings

SP - 1

EP - 4

BT - Speech Communication; 10. ITG Symposium

PB - IEEE

T2 - 10th ITG Symposium on Speech Communication

Y2 - 26 September 2012 through 28 September 2012

ER -

Dereverberation with an Iterative Least-Squares Technique and Minimum Mean-Square Error Estimation for Automatic Speech Recognition

Abstract

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this