Maximum likelihood PSD estimation for speech enhancement in reverberant and noisy conditions

Adam Kuklasinski, Simon Doclo, Jesper Jensen

Publikation: Bidrag til bog/antologi/rapport/konference proceedingKonferenceartikel i proceedingForskningpeer review

2 Citationer (Scopus)

Resumé

We propose a novel Power Spectral Density (PSD) estimator for multi-microphone systems operating in reverberant and noisy conditions. The estimator is derived using the maximum likelihood approach and is based on a blocked and pre-whitened additive signal model. The intended application of the estimator is in speech enhancement algorithms, such as the Multi-channel Wiener Filter (MWF) and the Minimum Variance Distortionless Response
(MVDR) beamformer. We evaluate these two algorithms in a speech dereverberation task and compare the performance obtained using the proposed and a competing PSD estimator. Instrumental performance measures indicate an advantage of the proposed estimator over the competing one. In a speech intelligibility test all algorithms significantly improved the word intelligibility score. While the results suggest a minor advantage of using the proposed PSD estimator, the difference between algorithms was found to be statistically significant only in some of the experimental conditions.
OriginalsprogEngelsk
TitelIEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016
ForlagIEEE
Publikationsdato25 mar. 2016
Sider599 - 603
ISBN (Elektronisk)978-1-4799-9988-0
DOI
StatusUdgivet - 25 mar. 2016
BegivenhedThe 41st IEEE International Conference on Acoustics, Speech and Signal Processing - Shanghai, Kina
Varighed: 20 mar. 201625 mar. 2016
http://www.icassp2016.org/

Konference

KonferenceThe 41st IEEE International Conference on Acoustics, Speech and Signal Processing
LandKina
ByShanghai
Periode20/03/201625/03/2016
Internetadresse

Fingerprint

Speech enhancement
Power spectral density
Maximum likelihood
Speech intelligibility
Microphones

Citer dette

Kuklasinski, A., Doclo, S., & Jensen, J. (2016). Maximum likelihood PSD estimation for speech enhancement in reverberant and noisy conditions. I IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016 (s. 599 - 603). IEEE. https://doi.org/10.1109/ICASSP.2016.7471745
Kuklasinski, Adam ; Doclo, Simon ; Jensen, Jesper. / Maximum likelihood PSD estimation for speech enhancement in reverberant and noisy conditions. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016. IEEE, 2016. s. 599 - 603
@inproceedings{1c2ab6b5a89d46c5b04d10f5dfd8987a,
title = "Maximum likelihood PSD estimation for speech enhancement in reverberant and noisy conditions",
abstract = "We propose a novel Power Spectral Density (PSD) estimator for multi-microphone systems operating in reverberant and noisy conditions. The estimator is derived using the maximum likelihood approach and is based on a blocked and pre-whitened additive signal model. The intended application of the estimator is in speech enhancement algorithms, such as the Multi-channel Wiener Filter (MWF) and the Minimum Variance Distortionless Response(MVDR) beamformer. We evaluate these two algorithms in a speech dereverberation task and compare the performance obtained using the proposed and a competing PSD estimator. Instrumental performance measures indicate an advantage of the proposed estimator over the competing one. In a speech intelligibility test all algorithms significantly improved the word intelligibility score. While the results suggest a minor advantage of using the proposed PSD estimator, the difference between algorithms was found to be statistically significant only in some of the experimental conditions.",
author = "Adam Kuklasinski and Simon Doclo and Jesper Jensen",
year = "2016",
month = "3",
day = "25",
doi = "10.1109/ICASSP.2016.7471745",
language = "English",
pages = "599 -- 603",
booktitle = "IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016",
publisher = "IEEE",
address = "United States",

}

Kuklasinski, A, Doclo, S & Jensen, J 2016, Maximum likelihood PSD estimation for speech enhancement in reverberant and noisy conditions. i IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016. IEEE, s. 599 - 603, The 41st IEEE International Conference on Acoustics, Speech and Signal Processing, Shanghai, Kina, 20/03/2016. https://doi.org/10.1109/ICASSP.2016.7471745

Maximum likelihood PSD estimation for speech enhancement in reverberant and noisy conditions. / Kuklasinski, Adam; Doclo, Simon; Jensen, Jesper.

IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016. IEEE, 2016. s. 599 - 603.

Publikation: Bidrag til bog/antologi/rapport/konference proceedingKonferenceartikel i proceedingForskningpeer review

TY - GEN

T1 - Maximum likelihood PSD estimation for speech enhancement in reverberant and noisy conditions

AU - Kuklasinski, Adam

AU - Doclo, Simon

AU - Jensen, Jesper

PY - 2016/3/25

Y1 - 2016/3/25

N2 - We propose a novel Power Spectral Density (PSD) estimator for multi-microphone systems operating in reverberant and noisy conditions. The estimator is derived using the maximum likelihood approach and is based on a blocked and pre-whitened additive signal model. The intended application of the estimator is in speech enhancement algorithms, such as the Multi-channel Wiener Filter (MWF) and the Minimum Variance Distortionless Response(MVDR) beamformer. We evaluate these two algorithms in a speech dereverberation task and compare the performance obtained using the proposed and a competing PSD estimator. Instrumental performance measures indicate an advantage of the proposed estimator over the competing one. In a speech intelligibility test all algorithms significantly improved the word intelligibility score. While the results suggest a minor advantage of using the proposed PSD estimator, the difference between algorithms was found to be statistically significant only in some of the experimental conditions.

AB - We propose a novel Power Spectral Density (PSD) estimator for multi-microphone systems operating in reverberant and noisy conditions. The estimator is derived using the maximum likelihood approach and is based on a blocked and pre-whitened additive signal model. The intended application of the estimator is in speech enhancement algorithms, such as the Multi-channel Wiener Filter (MWF) and the Minimum Variance Distortionless Response(MVDR) beamformer. We evaluate these two algorithms in a speech dereverberation task and compare the performance obtained using the proposed and a competing PSD estimator. Instrumental performance measures indicate an advantage of the proposed estimator over the competing one. In a speech intelligibility test all algorithms significantly improved the word intelligibility score. While the results suggest a minor advantage of using the proposed PSD estimator, the difference between algorithms was found to be statistically significant only in some of the experimental conditions.

U2 - 10.1109/ICASSP.2016.7471745

DO - 10.1109/ICASSP.2016.7471745

M3 - Article in proceeding

SP - 599

EP - 603

BT - IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016

PB - IEEE

ER -

Kuklasinski A, Doclo S, Jensen J. Maximum likelihood PSD estimation for speech enhancement in reverberant and noisy conditions. I IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016. IEEE. 2016. s. 599 - 603 https://doi.org/10.1109/ICASSP.2016.7471745