%% Project of EEL6586 --- LPC

clear;
%close all;
trainData = load('trainTemplate.mat');
trainNames = fieldnames(trainData);


N = 10;
windowSize = 200;
overlap = 100;
hamWindow = 0.54 - 0.46*cos(2*pi*(0:windowSize-1)/windowSize);



for s = 1 : length(trainNames) %% s is the number of sentences / instructions
    name = trainNames{s};
    data = getfield(trainData,name);
    
    figure(s);
    %% noise reduction
    for m = 1 : size(data,1) %% m is number of training data for each sentence
        %% pre-emphasis
        preData(m,:) = filter([1 -.97], 1, data(m,:));
        drvData(m,:) = conv2(preData(m,:), [0.333 0.666 0 -0.666 -0.333], 'same');
        winData(m,:) = wiener2(drvData(m,:));
        
        %% use end-point to delete silence
        Qs = filter(hamWindow,1, winData(m,:).* winData(m,:)*100);
        [dfoo, rang] = find(Qs >= 0.005);
        
        if rang(301) - rang(1) == 300
            rang1 = rang(1);
        else
            mm = 1;
            while rang(mm+1) - rang(mm) ==1 & rang(mm+300) - rang(mm) >300
                mm=mm+1;
            end
            rang1 = rang(mm+300);
        end
            
        if rang(length(rang)) - rang(length(rang)-300) == 300;
            rang2 = rang(length(rang));
        else
            mm = 0;
            while rang(length(rang)-mm) - rang(length(rang)-mm-1) ==1 & rang(length(rang)-mm) - rang(length(rang)-mm-300) > 300 
                mm = mm + 1;
            end
            rang2 = rang(length(rang)-mm-1);
        end
        
        trunkData = winData(m,rang1 : rang2);
        
        
        %% add windows to each training data and calculate LPC for each window
        %LPCtemp = [];
        LPCtemp = zeros(250,11);
        for n = 1 : (windowSize-overlap) : floor(length(trunkData) / windowSize-1) * windowSize
            windowed = trunkData(n : n + windowSize - 1);
            windowed = windowed .* hamWindow;
            LPCtemp((n-1) / (windowSize-overlap) + 1,:) = real(lpc(windowed, N));
        end
        LPCtemp(:,1) = []; 
        Qlpc = [];
        Qlpc(:,1) = filter(hamming(10), 1, LPCtemp(:,3));
        %Qlpc(:,1) = sign(Qlpc(:,1)-10)+1;
        Qlpc(:,2) = filter(hamming(10), 1, LPCtemp(:,4));
        %Qlpc(:,2) = sign(Qlpc(:,2)-10)+10;
        %sumlpc(s,m) = sum(Qlpc(:,2));
        %% calculate autocorrelation of each training data's LPC coeff.
        
        %R = LPCtemp * LPCtemp';
        %for k = 1 : N  %% k represents # of LPC autocorrelation
        %    r(s,m,k) = trace(R);
        %    R(1,:) = [];
        %    R(:,N-k+1) = [];
        %end
        
        %subplot(5,6,m);
        
        lpc_vector(s,m,:) = LPCtemp(:,4);
        
        %plot(LPCtemp(:,3:4));
        
        %hold on;
        %plot(Qlpc,'k');
        %axis tight;
        
        %r(s,m,:) = sum (LPCtemp(:,3:4))/ size(LPCtemp,1);
        
    end

 %   aMiddle(s,:) = mean(squeeze(r(s,:,:)));
 %   aLeft(s,:) = min(squeeze(r(s,:,:)));
 %   aRight(s,:) = max(squeeze(r(s,:,:)));    
        
end

%% test each training data
%for s = 1 : size(r,1)
%    for m = 1 : size(r,2)
        
%        for ss = 1 : size(aMiddle,1)
%            for k = 1 : size(r,3)
%                if r(s,m,k) <= aMiddle(ss,k) & r(s,m,k) >= aLeft(ss,k)
%                    b(ss,k) = (r(s,m,k) - aLeft(ss,k)) / ...
%                        (aMiddle(ss,k) - aLeft(ss,k)) * 0.2 + 0.8;
%                elseif r(s,m,k) > aMiddle(ss,k) & r(s,m,k) <=aRight(ss,k)
%                    b(ss,k) = (aRight(ss,k) - r(s,m,k)) / ...
%                        (aRight(ss,k) - aMiddle(ss,k)) * 0.2 + 0.8;
%                else b(ss,k) = 0;
%                end
%            end
%        end
%        B = sum(b,2);
%        [foo1,foo2] = max(B);
%        trainingResult(s,m) = foo2;
        
%    end
%end


%for s = 1 : size(lpc_vector,1) %% =6
%    for m = 1 : size(lpc_vector,2) %% =30
%        for ss = 1 : size(lpc_vector,1) %% =6
            
%            R = squeeze(lpc_vector(s,m,:))' * squeeze(lpc_vector(ss,:,:))';
            %if s == ss R(m) = 0; end
            %r(ss) = max(R);
            %if s == ss 
            %    r(ss) = sum(R)/29;
            %else
            %    r(ss) = mean(R);
            %       end
%        [foo(s,m) result(s,m)] = max(r);
%    end
%end
%sum_xcorr = 0;

for s = 1 : 6
    for m = 1:30
        for ss = 1:6
            for mm = 1:30
                R = xcorr(squeeze(lpc_vector(s,m,:)),squeeze(lpc_vector(ss,mm,:)));
                [max1 index] = max(R);
                max2 = max(R(index+15:499));
                max3 = max(R(1:index-15));
                if max2>max3
                    difference(mm) = max1 - max2;
                else
                    difference(mm) = max1 - max3;
                end
                
                %sum_xcorr = sum_xcorr + diff;
            end
            pp(ss) = sum(difference);
            
            %sum_xcorr = 0;
        end
        [foo(s,m) result(s,m)] = max(pp);
        
    end
end