function y = block_filter (X,B); 
% function y = block_filter (X,B); 
%
% block of filters with just one output. 
% X   - matrix of signals to filter, size P x N
% B   - matrix of numerator coeffs (in lines), but LR-flipped P x L 
%       (L is the length of filters). Actually, the coeffs are in order
%       to do rather a correlation than filtering !
%
% initial conditions are zero. 

[P,N] = size (X);  [P,L] = size (B); 
% prolong X with L-1 zeros
X = [zeros(P,L-1) X]; 

% init the output
y = zeros (1,N); 

for (ii=1:N)
  y(ii) = sum(sum(B .* X(:,ii:(ii+L-1))));
end