Why does the num channel and sample exchange position?

I run the following code

f=wavread('tes.wav','native');
v=[1/sqrt(2) 1/sqrt(2)]; 
w=[1/sqrt(2) -1/sqrt(2)]; 
if mod(length(f),2)~=0
    f=[f 0];
end
d=length(f);
m=1:d/2;
a1=f(2*m-1).*v(1) + f(2*m).*v(2);
d1=f(2*m-1).*w(1) + f(2*m).*w(2);

But something seems not right to me, the result:

the raw signal was = < 250608x2 int16 > but after i process it become like a1 < 1x125304 int16 > and d1 < 1x125304 int16 >, what i expected is like a1 < 125304x1 int16 > ? do you know how i can achieve this?

Why is that change from 250608x2 become 1x125304 anyway. m x n, m = number of sample and n=number of chanel right? why does it's exchange position?