2018
DOI: 10.1016/j.csl.2017.11.003
|View full text |Cite
|
Sign up to set email alerts
|

Rank-1 constrained Multichannel Wiener Filter for speech recognition in noisy environments

Abstract: Multichannel linear filters, such as the Multichannel Wiener Filter (MWF) andthe Generalized Eigenvalue (GEV) beamformer are popular signal processing techniques which can improve speech recognition performance. In this paper, we present an experimental study on these linear filters in a specific speech recognition task, namely the CHiME-4 challenge, which features real recordings in multiple noisy environments. Specifically, the rank-1 MWF is employed for noise reduction and a new constant residual noise powe… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
36
0

Year Published

2018
2018
2023
2023

Publication Types

Select...
4
3
1

Relationship

2
6

Authors

Journals

citations
Cited by 33 publications
(36 citation statements)
references
References 33 publications
0
36
0
Order By: Relevance
“…Originally proposed in [18] for multi-channel Wiener filtering and later applied to robust ASR [19], [20], the speech covariance matrix can be approximated using the decomposition technique:…”
Section: Rank-1 Approximation Of Speech Covariance Matrixmentioning
confidence: 99%
“…Originally proposed in [18] for multi-channel Wiener filtering and later applied to robust ASR [19], [20], the speech covariance matrix can be approximated using the decomposition technique:…”
Section: Rank-1 Approximation Of Speech Covariance Matrixmentioning
confidence: 99%
“…where α is a forgetting factor and H denotes Hermitian transposition. Similarly, the noise covariance matrix Σ n , which includes the statistics corresponding to the other speaker and background noise, can be estimated as The first channel of c j (t, f ) is then estimated via the rank-1 constrained multichannel Wiener filter (R1-MWF) [15], which imposes a rank-1 constraint on Σ j (t, f ):…”
Section: Adaptive Beamforming To Extract the Source Signalsmentioning
confidence: 99%
“…Wang et al [14] studied several variants of the MWF for ASR purposes. The most promising one is derived from a rank-1 approximation of the MWF based on the GEVD.…”
Section: Mutichannel Wiener Filtersmentioning
confidence: 99%
“…Recent approaches that derive DNN-based multichannel filters have proven very promising [7,13]. These include various beamformers derived from the speech and noise covariance matrices computed from the output mask [7,14] or an MWF derived by expectation-maximization [13]. Yet these approaches target a single speaker in a noisy environment while real-world scenarios often involve several speakers.…”
Section: Introductionmentioning
confidence: 99%