How to do embedding of watermark using PN sequence in audio signal using frequency hopping technique ?

For my project I am refering to the paper in which these lines are mentioned:-

Watermark embedded into the audio signal in frequency domain is the pseudorandom noise sequence. Depending on which bit is to be embedded the process varies :– if bit 1 is to be embedded the value of the coefficient with lower frequency is set to be k dB above the mean value, while the value of the other coefficient is set to be k dB below the mean value. If bit 0 is to be embedded the operation is done vice versa. The fact that mean value of each sub-band remains unchanged is used when removing the watermark.

For this I have written a code but I am not getting the unchanged mean. mean is getting changed. what should be done for getting unchanged mean.

• [f Fs]=wavread('timbaland.wav');
• N = length(f);
• slength = N/Fs;
• nf=1024; %number of point in DTFT
• Y = fft(f,nf);
• x = randn(1, N) + cos(2*pi*100/1024*(0:N-1));
• x(1024*ceil(end/1024)) = 0;
• z = reshape(x, 1024, []);
• y = 20*log10(abs(fft(z)));
• one_count = 0;
• zero_count = 0;
• G=1023; % Code length
• sd1 =randsrc(1,10,[0 1]); % Initial state of Shift register
• PN1=[];
• j=0;
• sd(1)=0;
• % First m-sequence
• for j=1:G
• PN1=[PN1 sd1(5)];
• if sd1(1)==sd1(4)
• temp1=0;
• zero_count = minus_count + 1;
• else temp1=1;
• one_count = one_count + 1;
• end
• sd1(1)=sd1(2);
• sd1(2)=sd1(3);
• sd1(3)=sd1(4);
• sd1(4)=sd1(5);
• sd1(5)=sd1(6);
• sd1(6)=sd1(7);
• sd1(7)=sd1(8);
• sd1(8)=sd1(9);
• sd1(9)=sd1(10);
• sd1(10)=temp1;
• %disp(j);
• %disp(sd1);
• end
• %disp(one_count);
• %disp(zero_count);
• g=autocorr(PN1);
• %key
• key=PN1;
• [m,n]=size(y);
• vks = [];
• for i=1:1
• temp= y(:,i);
• mean1= mean(temp);
• %disp(mean1);
• low = 0; high = n;
• index1 = low + (high-low) * rand;
• index1 = round(index1);
• if index1==0
• index1 = 1;
• end
• %fprintf('Index1 = %d',index1);
• %disp(strcat('Index1 = ', num2str(index1)));
• fv= temp(index1,1);
• low = 0; high = n;
• index2 = low + (high-low) * rand;
• index2 = round(index2);
• %fprintf('Index2 = %d',index2);
• if index2==0
• index2 = 1;
• end
• fsv= temp(index2,1);
• diff= mean1-20;
• K= (diff/mean1)*100;
• lf = 0;
• hf = 0;
• if (fv < mean1)
• lf = fv;
• hf = fsv;
• loc1 = index1;
• loc2 = index2;
• else
• lf = fsv;
• hf = fv;
• loc1 = index2;
• loc2 = index1;
• end
• addi = 0;
• subst = 0;
• for j=1:length(key)
• if(key(j) == 1)
• disp('Bit is 1');
• lf = mean1 + K;
• hf = mean1 - K;
• addi = addi + lf;
• subst = subst + hf;
• end
• if(key(j) == 0)
• disp('Bit is 0');
• lf = mean1 - K;
• hf = mean1 + K;
• addi = addi + hf;
• subst = subst + lf;
• end
• temp(loc1) = lf;
• temp(loc2) = hf;
• end
• mean2= mean(temp);
• %disp(mean2);
• vks(:,i) = temp;
• end

please tell me where i have done the mistake in this code. and also tell me that after doing fft only we get the coefficients ? Please reply as earliest possible. Thanks in advance. Link for whole document from where above lines are taken : Frequency Hopping Method for Audio Watermarking

Answer 1

[korada pavankumar]

• zero_count = minus_count + 1;

In this step , u are using minus count but in ur code u r not assigning any value so instead of minus count use zero count u get output.

• zero_count = zero_count + 1;