Finite Impulse Response Filter Design - Digital Signal Processing: Fundamentals and Applications - page 251

0

-50

-100

0 500 1000 1500 2000 2500 3000 3500 4000

Frequency (Hertz)

0

-500

-1000

-1500

-2000

-2500

0 500 1000 1500 2000 2500 3000 3500 4000

Frequency (Hertz)

FIGURE 7.20

Frequency responses of the designed bandstop filter using the Blackman window.

We choose N ¼ 35, an odd number. The normalized lower and upper cutoff frequencies are calculated as
U L ¼ 2p 1; 250
8; 000

¼ 0:3125p radians

U H ¼ 2p 2; 850
8; 000

¼ 0:7125p radians

and N ¼ 2M þ 1 ¼ 35. Using MATLAB, the design results are demonstrated in Program 7.6.
Program 7.6. MATLAB program for Example 7.11.
% Figure 7.20 (Example 7.11)
% MATLAB program to create Figure 7.20
%
N ¼ 35; Ftype ¼ 4; WnL ¼ 0.3125*pi; WnH ¼ 0.7125*pi; Wtype ¼ 5;fs ¼ 8000;
Bblack ¼ firwd(N,Ftype,WnL,WnH,Wtype);
freqz(Bblack,1,512,fs);
axis([0 fs/2 -120 10]);

Figure 7.20 shows the plot of the frequency responses of the designed bandstop filter. The designed filter
coefficients are listed in Table 7.10 .
Comparisons of filtering effects are illustrated in Figures 7.21A and 7.21B .In Figure 7.21A , the original
speech and speech processed by the bandstop filter are plotted. The processed speech contains most of the energy
of the original speech because most of the energy of the speech signal exists in the low-frequency band.
Figure 7.21B verifies the filtering frequency effects. The requency components ranging from 2,000 Hz to 2,200
Hz have been completely removed.