scispace - formally typeset
Open AccessJournal ArticleDOI

Microphone array post-filter based on noise field coherence

Iain McCowan, +1 more
- 01 Nov 2003 - 
- Vol. 11, Iss: 6, pp 709-716
TLDR
A more general expression of the post-filter estimation is developed based on an assumed knowledge of the complex coherence of the noise field that can be used to construct a more appropriate post- filter in a variety of different noise fields.
Abstract
This paper introduces a novel technique for estimating the signal power spectral density to be used in the transfer function of a microphone array post-filter. The technique is a generalization of the existing Zelinski post-filter, which uses the auto- and cross-spectral densities of the array inputs to estimate the signal and noise spectral densities. The Zelinski technique, however, assumes zero cross-correlation between the noise on different sensors. This assumption is inaccurate, particularly at low frequencies and for arrays with closely spaced sensors, and thus the corresponding post-filter is suboptimal in realistic noise conditions. In this paper, a more general expression of the post-filter estimation is developed based on an assumed knowledge of the complex coherence of the noise field. This general expression can be used to construct a more appropriate post-filter in a variety of different noise fields. In experiments using real noise recordings from a computer office, the modified post-filter results in significant improvement in terms of objective speech quality measures and speech recognition performance using a diffuse noise model.

read more

Content maybe subject to copyright    Report

ESEARCHR EPR ORTIDIAP
  
  

 



 


  



  
!"



 
   
 
  
 

 
 
       
!"# $ % &'()
*' + )
,)-
./0 0


    
     
 
    
 

           
            
            
              
           
            
           
             
              
            
   Æ       
  !           
 

 
 
                  
            
           
             
           
 
!"          # $#% 
       & ' (  $&'(%
   )        
      * ( $*(%   &'(  
 
)        )     
Æ              
              
               
     
 
"      
 + ," -        
 
)  +          
  -     .     /
        -      
            
              
 
           +  
             0
        Æ      
     $/%       
             
     1 2 3" 4      
           
   /         
 +          
              
     +        
  ,         /  
        0       
         +  -  1 
              
  Æ   -         
          5 
    
  
  
6            
            
   $     %
¼
7
8
¼
$!%

 
,
Time Alignment
Wiener
Post-filter
Estimation
9 !: 9   

   
       
7

$%

¼
       
¼
7
¼
¼

¼
$,%

      
;   
 
!"       
 # $#%   $    ) %   
 )         & '
( $&'(%   

7


8

<

<

$1%


      Æ



   $%
        <

  $%   
           ) 
            &'(  2"
              9 ! 
       )      
             
           * ( $*(% 
            
7


8

$2%
  
       +       
 ,"      
 
"    
          9 !
-                
    /    
      
     
7
8
$3%

           

Citations
More filters
Journal ArticleDOI

A Consolidated Perspective on Multimicrophone Speech Enhancement and Source Separation

TL;DR: This paper proposes to analyze a large number of established and recent techniques according to four transverse axes: 1) the acoustic impulse response model, 2) the spatial filter design criterion, 3) the parameter estimation algorithm, and 4) optional postfiltering.
Journal ArticleDOI

Deep Learning for Environmentally Robust Speech Recognition: An Overview of Recent Developments

TL;DR: A review of recently developed, representative deep learning approaches for tackling non-stationary additive and convolutional degradation of speech with the aim of providing guidelines for those involved in the development of environmentally robust speech recognition systems.
Book ChapterDOI

Adaptive Beamforming and Postfiltering

TL;DR: This chapter explores many of the basic concepts of array processing with an emphasis on adaptive beamforming for speech enhancement applications and derives the frequency-domain linearly constrained minimum-variance beamformer, and its generalized sidelobe canceller (GSC) variant.
Journal ArticleDOI

A Multimodal Approach to Blind Source Separation of Moving Sources

TL;DR: Experimental results confirm that by utilizing the visual modality, the proposed algorithm improves the performance of the BSS algorithm and mitigates the permutation problem for stationary sources, but also provides a good BSS performance for moving sources in a low reverberant environment.
Journal ArticleDOI

Model-Based Dereverberation Preserving Binaural Cues

TL;DR: This contribution presents a novel two-stage bINAural dereverberation algorithm which explicitly preserves the binaural cues and significantly improves speech quality according to objective and subjective measures.
References
More filters
Journal ArticleDOI

Robust adaptive beamforming

TL;DR: It is shown that a simple scaling of the projection of tentative weights, in the subspace orthogonal to the linear constraints, can be used to satisfy the quadratic inequality constraint.
Proceedings ArticleDOI

A microphone array with adaptive post-filtering for noise reduction in reverberant rooms

R. Zelinski
TL;DR: The author presents a self-adapting noise reduction system which is based on a four-microphone array combined with an adaptive postfiltering scheme which produces an enhanced speech signal with barely noticeable residual noise if the input SNR is greater than 0 dB.
Journal ArticleDOI

Measurement of Correlation Coefficients in Reverberant Sound Fields

TL;DR: In this paper, the cross-correlation coefficient for the sound pressure at two points a distance r apart is defined and an instrument for measuring and recording R as a function of time is described.
Related Papers (5)