vqlbg.m

function c = vqlbg(d, k)
% VQLBG Vector quantization using the Linde-Buzo-Gray algorithm
%
% Inputs:
%       d contains training data vectors (one per column)
%       k is number of centroids required
%
% Outputs:
%       c contains the result VQ codebook (k columns, one for each centroids)

e = 0.01;
% dist_thresh = 0.1;
% dist_thresh = 3.1;
dist_thresh = 0.1;
q=1;
%computes the initial centroid of all the acoustic vectors
centr = mean(d,2);
codebook = centr;

while  size(codebook,2)<k
    
split = zeros(size(d , 1) , size(codebook,2)*2);
for i = 1 : size(codebook,2)
   %splits the codebook; for each codeword currently in the book, it splits
   %it into two (doubling the size of the codebook)
%     split(: , 2*i-1:2*i ) = [codebook(:,i)-e , codebook(:,i)+e];
    split(: , 2*i-1:2*i ) = [codebook(:,i).*(1+e) , codebook(:,i).*(1-e)];


end
    codebook = split;
    convg = 10;
    dist = 10;
    while convg > dist_thresh
        z = disteu(d , codebook);
        %finds the closest codeword in codebook1 to each acoustic vector, where
        %the vector output, j=ind(i) tells that acoustic vector i is closes to
        %codework j
        [m , ind] = min(z , [] , 2);
        
        convg = abs(mean(m) - dist); 
        dist = mean(m);
        indices = unique(ind);

        %finds all the acoustic vectors associated with code word j, and thus
        %are all part of the same cluster
        codebook = zeros(size(d , 1) , size(indices,1));
        for j = 1 : size(indices,1)
           %seperates the clusters and find the new centroids for the updates
           %the code book
           clust = d(:, find(ind == indices(j) ));
           codebook(: , j) = mean(clust,2); 

%            if avg_dist < dist_thresh
% %                 subplot(2,2,q)
% % %                 q=q+1;
% %                 plot(clust(1,:), clust(2,:),'o')
% %                 hold on
% %                 plot(codebook(1,j), codebook(2,j),'o','LineWidth',3)
%            end
        end
    end 
   q;
   q=q+1;
end
c = codebook;
% figure(q-1)
% plot(d(1,:),d(2,:),'o')
% hold on 
% plot(c(1,:) , c(2,:), 'o','LineWidth',3)

end