Page 98 - Algorithm Collections for Digital Signal Processing Applications using MATLAB

86 Chapter 2

Note that the signal at the top and the bottom of the figure 2-7 are almost
identical. This preprocessing stage is normally used before extracting
features from the signal for pattern recognition.

5.3 M-program for Mean and Variance Normalization

________________________________________________________
meanvarnormgv.m

%a is the speech signal to be modified
%b is the reference signal with desired mean and variance.
% res is the normalilzed version of signal ‘a’ with desired mean and variance
m=mean(a);
v=sqrt(var(a));
md=mean(b);
vd=sqrt(var(b));
delta=abs((a-m)*vd/v);
res=ones(1,length(a))*md;
[p]=quantiz(a,m);
p=p'; p=p*2-1;
res=res+p.*delta;
________________________________________________________________________