Digital Signal Processing Reference
In-Depth Information
We can nish the reconstruction for the second octave by looking at the result when
w2
f
[n] and z2
f
[n] are added together to recreate w
d
[n]. We will call this recreated
signal w
d
[n]
0
, until we are certain that it is the same as w
d
[n].
0
w
d
[n]
= w2
f
[n] + z2
f
[n]
0
w
d
[n]
= dw2[n] + bw2[n2]az2[n]cz2[n2]; n is even-even
0
w
d
[n]
= cw2[n1] + aw2[n3] + bz2[n1] + dz2[n3]; n is odd-even
The important thing to do here is to show that the two signals marked w
d
[n] in
Figure 9.22 are exactly the same. This means nding the previous expression in
terms of w[n] only. First, we will take the expressions for w2[n] and z2[n], and nd
w2[nk] and z2[nk].
w2[n] = aw
d
[n] + bw
d
[n1] + cw
d
[n2] + dw
d
[n3]
z2[n] = dw
d
[n]cw
d
[n1] + bw
d
[n2]aw
d
[n3]
w2[nk] = aw
d
[nk] + bw
d
[nk1] + cw
d
[nk2] + dw
d
[nk3]
z2[nk] = dw
d
[nk]cw
d
[nk1] + bw
d
[nk2]aw
d
[nk3]
Now, we can plug these into the previous w
d
[n] expressions.
0
w
d
[n]
= d(aw
d
[n] + bw
d
[n1] + cw
d
[n2] + dw
d
[n3])
+ b(aw
d
[n2] + bw
d
[n3] + cw
d
[n4] + dw
d
[n5])
a(dw
d
[n]cw
d
[n1] + bw
d
[n2]aw
d
[n3])
c(dw
d
[n2]cw
d
[n3] + bw
d
[n4]aw
d
[n5]); n is even-even
0
w
d
[n]
= c(aw
d
[n1] + bw
d
[n2] + cw
d
[n3] + dw
d
[n4])
+ a(aw
d
[n3] + bw
d
[n4] + cw
d
[n5] + dw
d
[n6])
+ b(dw
d
[n1]cw
d
[n2] + bw
d
[n3]aw
d
[n4])
+ d(dw
d
[n3]cw
d
[n4] + bw
d
[n5]aw
d
[n6]); n is odd-even
Rewriting these expressions and lining them up into columns gives us: