getClassificationLRT_M.m

function Res  = getClassificationLRT_M(noise2, clean2 , GT, eta, m)
clean1 = exp(-clean2);
noise1 = exp(-noise2);
lrt = log(clean1./noise1);
[o1 o2 o3] = size(clean2);
for j = 1 : o1; 
    if m > 0
        for j1 = 1 : 50;
            R2 = [ zeros(1,m) squeeze(lrt(j,j1,:))' zeros(1,m)];
            R2 = sum(buffer(R2, 2*m+1,2*m, 'nodelay'));
            R1(1,j1,:) = R2';
        end
    
    else 
           R1 = lrt(j,:,:);
    end
     
    corr = sum(sum(GT(j,:,1:237)));
	wrong = sum(sum(GT(j,:, 1:237) == 0));
	R = zeros(size(R1));
	R(R1 >= eta) = 1;
	R(R1 < eta) = 0;
%	max(max(R1))
%	min(min(R1))
	%%%% Correct %%%
	Res.acc(j) = sum(sum(R == GT(j,:,1:237)))*100/(50*237);  
	temp  = R.*GT(j,:,1:237);
	Res.S_S(j) = sum(sum(temp));
	Res.N_S(j) = (sum(sum(R)) - Res.S_S(j));
	temp1 = abs(R -1);
	temp2 = abs(GT(j,:,1:237) -1);
	temp3 = squeeze(temp1).*squeeze(temp2);
	Res.N_N(j) = sum(sum(temp3));
	Res.S_N(j) = (sum(sum(temp1)) - Res.N_N(j));
	Res.S_S(j) = Res.S_S(j)*100/corr;
	Res.S_N(j) = Res.S_N(j)*100/corr;
	Res.N_S(j) = Res.N_S(j)*100/wrong;	
	Res.N_N(j) = Res.N_N(j)*100/wrong;	
	clear temp*
	
end