Minimum Mean-Square Error Estimation of Discrete Fourier Coefficients With Generalized Gamma Priors

Erkelens, J.S.; Hendriks, Richard C.; Heusdens, Richard; Jensen, Jesper

doi:10.1109/tasl.2007.899233

Cited by 229 publications

(201 citation statements)

References 21 publications

Supporting

Mentioning

199

Contrasting

Order By: Relevance

“…The noise in the MVDR beamformer output remains Gaussian, since is assumed to be complex-Gaussian distributed and the MVDR beamformer is linear and deterministic. From (9) we can conclude that the MWF is suboptimal in general. An exception is the case in which , i.e., the complex speech DFT coefficients are assumed to have a complex-Gaussian distribution, for which (9) reduces to the MWF solution.…”

Section: A Assuming Complex-gaussian Noise Dft Coefficientsmentioning

confidence: 88%

“…Using Bayes rule, and the fact that the speech phase is uniformly distributed and independent of the speech magnitude [9], we obtain (3)…”

Section: Optimal Multichannel Mse Estimatorsmentioning

confidence: 99%

“…To specify the post-processing applied in (7) to , we must specify . As in [9] we assume that the magnitude is generalized-Gamma distributed, i.e., (8) where is the Gamma function and where we set from which it follows [9] that , with . Let denote the confluent hypergeometric function [18].…”

Section: A Assuming Complex-gaussian Noise Dft Coefficientsmentioning

confidence: 99%

“…Experimental Validation: To experimentally validate the suboptimality of the MWF we conduct an experiment where the performance of the multichannel MMSE estimator in (9) is compared to the MWF and the single-channel MMSE estimator presented in [19] that is derived under exactly the same distributional assumptions as the estimator in (9). The comparison is performed in terms of the quantity that the MMSE estimators optimize for, i.e., mean-squared error (MSE), estimated as (11) where and denote the total number of frequency bins and time-frames, respectively.…”

Section: A Assuming Complex-gaussian Noise Dft Coefficientsmentioning

confidence: 99%

“…Bayesian estimators [6], e.g., minimum mean-squared error (MMSE) estimators, have played an important role for single-channel methods, e.g., [1], [2], [7]- [9], as well as multichannel methods, e.g., [10]- [12]. A well-known multichannel estimator is the multichannel Wiener filter (MWF), e.g., [10], [13].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

On Optimal Multichannel Mean-Squared Error Estimators for Speech Enhancement

Hendriks

Heusdens

Kjems

et al. 2009

IEEE Signal Process. Lett.

View full text Add to dashboard Cite

Abstract-In this letter we present discrete Fourier transform (DFT) domain minimum mean-squared error (MMSE) estimators for multichannel noise reduction. The estimators are derived assuming that the clean speech magnitude DFT coefficients are generalized-Gamma distributed. We show that for Gaussian distributed noise DFT coefficients, the optimal filtering approach consists of a concatenation of a minimum variance distortionless response (MVDR) beamformer followed by well-known single-channel MMSE estimators. The multichannel Wiener filter follows as a special case of the presented MSE estimators and is in general suboptimal. For non-Gaussian distributed noise DFT coefficients the resulting spatial filter is in general nonlinear with respect to the noisy microphone signals and cannot be decomposed into an MVDR beamformer and a post-filter.

show abstract

Section: A Assuming Complex-gaussian Noise Dft Coefficientsmentioning

confidence: 88%

“…Using Bayes rule, and the fact that the speech phase is uniformly distributed and independent of the speech magnitude [9], we obtain (3)…”

Section: Optimal Multichannel Mse Estimatorsmentioning

confidence: 99%

Section: A Assuming Complex-gaussian Noise Dft Coefficientsmentioning

confidence: 99%

Section: A Assuming Complex-gaussian Noise Dft Coefficientsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

On Optimal Multichannel Mean-Squared Error Estimators for Speech Enhancement

Hendriks

Heusdens

Kjems

et al. 2009

IEEE Signal Process. Lett.

View full text Add to dashboard Cite

show abstract

Beamforming

2009

Distant Speech Recognition

View full text Add to dashboard Cite

A complete system for distant speech recognition (DSR) typically consists of several distinct components. While it is tempting to isolate and optimize each component individually, experience has proven that such an approach cannot lead to optimal performance. In this talk, we will discuss several examples of the interactions between the individual components of a DSR system. In addition, we will describe the synergies that become possible as soon as each component is no longer treated as a "black box". To wit, instead of treating each component as having solely an input and an output, it is necessary to peal back the lid and look inside. It is only then that it becomes apparent how the individual components of a DSR system can be viewed not as separate entities, but as the various organs of a complete body, and how optimal performance of such a system can be obtained.

show abstract

Voice Activity Detection, Noise Estimation, and Adaptive Filters for Acoustic Signal Enhancement

Martin¹,

Kolossa²

2012

Techniques for Noise Robustness in Automatic Speech Recognition

View full text Add to dashboard Cite

Minimum Mean-Square Error Estimation of Discrete Fourier Coefficients With Generalized Gamma Priors

Cited by 229 publications

References 21 publications

On Optimal Multichannel Mean-Squared Error Estimators for Speech Enhancement

On Optimal Multichannel Mean-Squared Error Estimators for Speech Enhancement

Beamforming

Voice Activity Detection, Noise Estimation, and Adaptive Filters for Acoustic Signal Enhancement

Contact Info

Product

Resources

About