I have written a code please help me in getting the output .

When i run this code i got a warninig that data cliped during write to file. Please sugggest something.

 clear;
  clc;
  [data,Fs]=wavread('bird');
  predictor = [0 1];                     %%y(k)=x(k-1)
  M=2;
  freq_sep= 100;
  nsamp=6;
  fs=1000;
  snr=0:1:25;
  for i=1:1:length(snr)
  partition = [-1 1];
  codebook = [-1 .50 1];

% Quantize x using DPCM.
encodedx = dpcmenco(data,codebook,partition,predictor);     %%modulated signal

fsk_mod=fskmod(encodedx,M,freq_sep,nsamp,fs);
noise=awgn(fsk_mod,snr(i));
fsk_demod=fskdemod(noise,M,freq_sep,nsamp,fs);
decodedx = dpcmdeco(fsk_demod,codebook,predictor);
[num,rate(i)]=biterr(encodedx,fsk_demod);
wavwrite(decodedx,Fs,8,'Bird')
sound('Bird')
end
semilogy(snr,rate)

[Daniel Shub]

While unrelated to the warning you are getting, I am pretty sure

sound('Bird')

is not valid MATLAB syntax

Answer 1

[Daniel Shub]

Wow, this question keeps coming up. The WAVWRITE function expects its input (in your case decodedx) to be an array with all values having a magnitude less than 1. You need to scale your decodedx.

See

http://www.mathworks.com/matlabcentral/answers/37866-having-errors-while-converting-mat-file-to-wav-file#answer_47369

Answer 2

[Geoff]

Is the output wave supposed to be 8-bit? That's exceptionally grungey... I don't use these wave file functions, but I expect that 'clipped' refers to the signal peaks getting chopped off due to a sampling issue. It could be that your source is 16-bit.

If you type:

class(decodedx)

Does it say uint8?