How to implement the Huffman codes to voice signal?

Hi,

I have a voice signal that I have quantized and now I have to employ Huffman coding to the quantized samples. I have tried to implement the "huffmanenco" but I keep getting the same error message, I would be glad if someone can help me out.

Here are the codes that I have written:

...
%Quantization
partition=[-0.2:0.04:0.2];
codebook=[-.24:0.04:-0.2];
[index,quants] = quantiz(myRecording,partition,codebook)
plot(quants)

%Probability distribution

   prob_y = arrayfun(@(x)length(find(quants==x)), unique(quants)) / length(quants)

%Huffman Encoding
symbols = [1:17]; % Distinct symbols that data source can produce
p = prob_y; % Probability distribution
[dict,avglen] = huffmandict(symbols,p); % Create dictionary.
comp = huffmanenco(quants,dict);

Error: The Huffman dictionary provided does not have the codes for all the input signals.

Thanks in advance

Answer 1

[Walter Roberson]

Look at the documentation again. http://www.mathworks.com/help/comm/ref/quantiz.html

quants is the floating point values, the codebook indexed by the appropriate index. If you want the symbol range to be 1:17 then you want to be passing in "index" not quants.

Answer 2

[aaa bbb]

Thanks a lot

[Parth Patel]

if u got the output now , kindly share the whole code for that