Digital Signal Processing Reference
In-Depth Information
where 1 ≤ α 1,1 < 1 and 0 ≤ β 1,1 1. We immediately find that the unique
solution to (3.34) is
1 − β 1,1 e ωτ 1
(1 − β 1,1 ) e
1
1 − e ωτ 0 (1 − α 1,1 )
h ( ω )=
,
(3.35)
ωτ 2
where τ 1 = τ 0 (1 − α 1,1 ) and τ 2 = τ 0 α 1,1 . Approximating 1 −e ωτ 0 (1 − α 1,1 )
with (3.6), we get
1 − β 1,1 e ωτ 1
(1 − β 1,1 ) e
C
1 − α 1,1
· 1
ω
h ( ω )
,
(3.36)
ωτ 2
from which we deduce the equivalent filter 1 :
1 − β 1,1 e ωτ 1
(1 − β 1,1 ) e
( ω )= 1
ω
h
.
(3.37)
ωτ 2
Taking the complex conjugate of the components of h
( ω ), we get the gains
that should be applied at the two microphone outputs:
ωτ 1
1 ( ω )= 1 − β 1,1 e
H ′∗
,
(3.38)
ω
2 ( ω )= (1 − β 1,1 ) e ωτ 2
ω
′∗
H
,
(3.39)
ωτ 1 and e ωτ 2 are time-domain delays τ 1 and −τ 2 at the
first and second microphone outputs, respectively. Figure 3.11 illustrates the
implementation of the first-order differential array.
The beampattern corresponding to the beamformer h
where the term e
( ω ) is
1 − β 1,1 e ωτ 1 (1 − β 1,1 ) e ω ( τ 0 cos θ − τ 2 )
( ω ) ]= 1
ω
B [ h
(3.40)
and applying the approximation (3.6) to the previous expression leads to
( ω ) ] 1 − α 1,1
C
1 1 − β 1,1
1 − α 1,1
+ 1 − β 1,1
B [ h
1 − α 1,1 cos θ
,
(3.41)
which is recognized to be the pattern of the first-order DMA. There are five
interesting cases.
Dipole: α 1,1 = β 1,1 = 0 (see Section 3.2).
Cardioid: α 1,1 = 1 1,1 = 0 (see Section 3.3).
Subcardioid [2], [3]: α 1,1 = 1 1,1 =0 . 4.
Hypercardioid: α 1,1 = 2 1,1 = 0.
1 Notice that this filter is noncausal for Α 1,1 > 0. In this case, we need to add a processing
delay equal to Τ 0 Α 1,1 = Τ 2 .
Search WWH ::




Custom Search