softdataassignmentweights.m

语音信号处理前分离信号噪声,matlab环境下运行,使信号处理效果更好

% SOFTDATAASSIGNMENTWEIGHTS soft-LOST soft data assignment
%
%  dataWeights = softDataAssignmentWeights(M,N,B,signalSize,data,lineVectors)
%
%  inputs;
%  M : Number of sources;
%  N : Number of mixtures;
%  B : Soft boundary value;
%  signalSize : Number of data points;
%  data : Mixture data;
%  lineVectors : Line orientation vectors
%
%  outputs:
%  dataWeights : Partial Assignment of data points to line orientations
%
%  
% Code available at http://www-bcl.cs.may.ie/~pogrady/
%
% Function provided as part of Soft-LOST matlab functions
%
% Paul D. O'Grady - 14-11-2003
% ([e-mail]) 


function dataWeights = softDataAssignmentWeights(M,N,B,signalSize,data,lineVectors)


null_vector = zeros(N,1);
zero_weights = zeros(1,M);

% defining size in advance makes code run quicker
Z(signalSize,M) = 0; 
 
for j = 1:signalSize

	dataPoint = data(:,j);%normalise(data(:,j));
		
	if dataPoint ~= null_vector
			
		for count = 1:M
			  
			%test the data with its oppsite signs
			R(count) = norm(dataPoint'-(lineVectors(count,:)*dataPoint)*lineVectors(count,:));
						
		end

		R = -B*R;
		expR = exp(R);
		expRSum = sum(expR);

		Z(j,:) = expR/expRSum;

	else
		Z(j,:) = zero_weights;

	end;


end;
dataWeights = Z;