LAB 13 : Noise Reduction

Contents
1. Noise Reduction
2. Optimal Filter Design
3. Butterworth Filter Design
4. Chebyshev Filter Design

LAB 13: Noise Reduction 2

 Noise Reduction
Introduction

 A digital filter removes noise in

– A signal that is contaminated by noise existing in a
broad frequency range
 The desired signal usually occupies a certain
frequency range
– Design a digital filter to remove frequency components
other than the desired frequency range

LAB 13: Noise Reduction 3

 Noise Reduction
Example1: Data Acquisition System (1)

 Record a 500 Hz sine wave with 𝑓𝑠 = 8 kHz.

 The signal is corrupted by broadband noise 𝜈 𝑛 :
𝑥 𝑛 = 1.4141 ∙ sin 2𝜋 ∙ 500𝑛/8,000 + 𝜈 𝑛
 Noise is broadband existing from 0 Hz to the
folding frequency of 4,000 Hz.
 Assuming that the desired signal has a frequency
range of only 0 to 800 Hz.
 Filter specifications
– Passband frequency range: 0 ~ 800 Hz with the
passband ripple less than 1 dB

LAB 13: Noise Reduction 4

 Noise Reduction
Example1: Data Acquisition System (2)

– Stopband frequency range: 2 ~ 4 kHz with 20 dB

attenuation
 Designed filter
– Rectangular window
– Cut-off frequency : 1400 Hz
𝑓stop −𝑓pass 2000−800 3
– ∆𝑓 = = =
𝑓𝑠 8000 20
0.9
– Filter length = =6𝑁=7
∆𝑓

LAB 13: Noise Reduction 5

 Noise Reduction
Example1: Data Acquisition System (3)

% MATLAB (Application of noise filtering)

close all; clear all
fs=8000; % sampling rate
T=1/fs; % sampling period
v=sqrt(0.1)*randn(1,250); % Gaussian random noise
n=0:1:249; % indexes
x=sqrt(2)*sin(2*pi*500*n*T)+v; % 500-Hz plus noise
figure(1)
subplot(2,1,1);plot(n,x);
xlabel('Number of samples');ylabel('Sample value');grid;
N=length(x);

LAB 13: Noise Reduction 6

 Noise Reduction
Example1: Data Acquisition System (4)

% Calculate single side spectrum for x(n)

f=[0:N/2]*fs/N;
Axk=2*abs(fft(x))/N;
Axk(1)=Axk(1)/2;
subplot(2,1,2);
plot(f,Axk(1:N/2+1));
xlabel('Frequency (Hz)');
ylabel('Amplitude |X(f)| ');grid;
figure(2)
Fc=1400/(fs/2); % normalized cut-off frequency
B=fir1(6,Fc,boxcar(7));% design FIR filter (filter length = 7)
y=filter(B,1,x); % perform digital filtering

LAB 13: Noise Reduction 7

 Noise Reduction
Example1: Data Acquisition System (5)

Ayk=2*abs(fft(y))/N;
Ayk(1)=Ayk(1)/2; % single-side spectrum of the filtered data
subplot(2,1,1); plot(n,y);
axis([0 length(v) -4 4]);
xlabel('Number of samples');
ylabel('Sample value');grid;
subplot(2,1,2);plot(f,Ayk(1:N/2+1));
axis([0 fs/2 0 1.5]);
xlabel('Frequency (Hz)');
ylabel('Amplitude |Y(f)| ');grid;

LAB 13: Noise Reduction 8

 Noise Reduction
Example1: Data Acquisition System (6)

LAB 13: Noise Reduction 9

 Noise Reduction
Example1: Data Acquisition System (7)

LAB 13: Noise Reduction 10

 Noise Reduction
Example2: Speech Recording System (1)

 Record speech in a noisy environment at a

sampling rate of 8000 Hz
 The recorded speech contains information within
1800 Hz
 Specification
– Filter type = lowpass FIR
– Passband frequency range = 0 – 1,500 Hz
– Passband ripple = 1 dB
– Stopband frequency range = 2500 – 4000 Hz
– Stopband attenuation = 30 dB

LAB 13: Noise Reduction 11

 Speech Noise Reduction
Example2: Speech Recording System (2)

 Filter design
– Window type = Hanning window
– Number of filter taps = 25
– Lowpass cutoff frequency = 2000 Hz

LAB 13: Noise Reduction 12

 Speech Noise Reduction
Example2: Speech Recording System (3)

% Ex2 - Speech Recording System

close all; clear all; clc
fs=8000;T=1/fs;
load speech.dat
we=speech;
t=[0:length(we)-1]*T;
th=mean(we.*we)/4;
v=sqrt(th)*randn([1,length(we)]);
x=we+v;
subplot(2,1,1);plot(t,x);
xlabel('Number of samples');ylabel('Sample value');grid;
title('Noisy Speech and its spectrum');

LAB 13: Noise Reduction 13

 Speech Noise Reduction
Example2: Speech Recording System (4)

N=length(x);
f=[0:N/2]*fs/N;
Axk=2*abs(fft(x))/N; Axk(1)=Axk(1)/2;
subplot(2,1,2); plot(f,Axk(1:N/2+1));
xlabel('Frequency (Hz)');
ylabel('Amplitude Spectrum |X(f)| '); grid;
figure
N1=24; % filter degree
fc=2000/4000; % normalized cutoff frequency of 2000 Hz
B=fir1(N1,fc,hanning(N1+1));
y=filter(B,1,x);
Ayk=2*abs(fft(y))/N;Ayk(1)=Ayk(1)/2;

LAB 13: Noise Reduction 14

 Speech Noise Reduction
Example2: Speech Recording System (5)

subplot(2,1,1); plot(t,y);
xlabel('Number of samples');
ylabel('Sample value');grid;
title('Enhanced Speech and its Spectrum');
subplot(2,1,2); plot(f,Ayk(1:N/2+1));
xlabel('Frequency (Hz)');
ylabel('Amplitude Spectrum |Y(f)| ');grid;
pause;
sound(x/max(abs(x)),fs);
pause
sound(y/max(abs(y)),fs);

LAB 13: Noise Reduction 15

 Speech Noise Reduction
Example2: Speech Recording System (6)

LAB 13: Noise Reduction 16

 Speech Noise Reduction
Example2: Speech Recording System (7)

LAB 13: Noise Reduction 17

 Noise Reduction
in Vibration Signals (1)
 Data acquisition system for vibration analysis
– Captured a vibration signal using an accelerometer
sensor in the noisy environment (𝑓𝑠 = 1,000 Hz)
– Vibration analysis requires the first dominant frequency
component in the range 35 to 50 Hz to be retrieved.
 Filter specifications
– Filter type: bandpass filter
– Passband frequency range = 35 ~ 50 Hz
– Passband ripple = 0.02 dB
– Stopband frequency ranges = 0~15 and 70~500 Hz
– Stopband attenuation = 50 dB

LAB 13: Noise Reduction 18

 Noise Reduction
in Vibration Signals (2)
 Filter design
– Window type = Hamming window
– Number of filter taps = 167
– Low cutoff frequency = 25 Hz
– High cutoff frequency = 65 Hz

LAB 13: Noise Reduction 19

 Noise Reduction
in Vibration Signals (3)

Noisy vibration signal and its spectrum

LAB 13: Noise Reduction 20

 Noise Reduction
in Vibration Signals (4)

Retrieved vibration signal and its spectrum

LAB 13: Noise Reduction 21

 firpmord
Syntax

 Parks-McClellan optimal FIR filter order

estimation
 Syntax
>> [n,fo,ao,w] = firpmord(f,a,dev)
>> [n,fo,ao,w] = firpmord(f,a,dev,fs)

LAB 13: Noise Reduction 22

 firpmord
Description

>> [n,fo,ao,w] = firpmord(f,a,dev)

– Returns the approximate order n, normalized frequency band
edges fo, frequency band amplitudes ao, and weights w that
meet input specifications f, a, and dev.

>> [n,fo,ao,w] = firpmord(f,a,dev,fs)

– Specifies a sampling frequency fs. fs defaults to 2 Hz, implying a
Nyquist frequency of 1 Hz. You can specify band edges scaled
to a particular application's sample rate.

LAB 13: Noise Reduction 23

 firpmord
Input Arguments (1)

 f  Frequency band edges, real-valued vector

– Frequency band edges, specified as a real-valued vector.
The argument must be in the range [0, Fs/2],
where Fs is the sampling frequency.
– The number of elements in the vector is always a
multiple of 2. The frequencies must be in increasing
order.
 a  Desired amplitude, vector
– Desired amplitudes at the points contained in f, specified
as a vector.
– f and a must satisfy the condition of length(f) =
2*length(a)–2. The desired function is piecewise
constant.
LAB 13: Noise Reduction 24
 firpmord
Input Arguments (2)

 dev — Maximum allowable deviation, vector

– Maximum allowable deviation, specified as a
vector. dev has the same size as a.
– It specifies the maximum allowable deviation or ripples
between the frequency response and the desired
amplitude of the output filter for each band.
 fs — Sample rate, 2 Hz (default) | real scalar
– Sample rate, specified as a real scalar.

LAB 13: Noise Reduction 25

 Firpmord
Output Arguments (1)

 n — Filter order, positive integer

– Filter order, returned as a positive integer.

 fo — Normalized frequency points, real-valued

vector
– Normalized frequency points, specified as a real-valued
vector.
– The argument must be in the range [0, 1], where 1
corresponds to the Nyquist frequency (Fs/2).
– The number of elements in the vector is always a
multiple of 2.
– The frequencies must be in increasing order.

LAB 13: Noise Reduction 26

 Optimal Filter Design
firpmord: Output Arguments (2)

 ao — Amplitude response, real-valued vector

– Amplitude response, returned as a real-valued vector.
 w — weights, real-valued vector
– Weights used to adjust the fit in each frequency band,
specified as a real-valued vector.
– The length of w is half the length of f, so there is exactly
one weight per band.

 In some cases, firpmord underestimates or

overestimates the order n. If the filter does not
meet the specifications, try a higher order such
as n+1 or n+2.

LAB 13: Noise Reduction 27

 Optimal Filter Design
firpmord: Example (1)

 Design a minimum-order lowpass filter with a

500 Hz passband cutoff frequency and 600 Hz
stopband cutoff frequency.
 Specify a sampling frequency of 2000 Hz.
 Require at least 40 dB of attenuation in the
stopband and less than 3 dB of ripple in the
passband.
>> rp = 3; % Passband ripple in dB
>> rs = 40; % Stopband ripple in dB
>> fs = 2000; % Sampling frequency
>> f = [500 600]; % Cutoff frequencies
>> a = [1 0]; % Desired amplitudes

LAB 13: Noise Reduction 28

 Optimal Filter Design
firpmord: Example (2)

 Convert the deviations to linear units.

 Design the filter and visualize its magnitude and
phase responses.
>> dev=[(10^(rp/20)-1)/(10^(rp/20)+1), 10^(-rs/20)];
>> [n,fo,ao,w] = firpmord(f,a,dev,fs);
>> b = firpm(n,fo,ao,w);
>> freqz(b,1,1024,fs)
>> title('Lowpass Filter Designed to Specifications')

LAB 13: Noise Reduction 29

Optimal Filter Design
firpmord: Example (3)

The filter falls slightly short

of meeting the stopband
attenuation and passband
ripple specifications.
Using n+1 instead of n in
the call to firpm achieves
the desired amplitude
characteristics.

LAB 13: Noise Reduction 30

 Butterworth Filter Design
BUTTER (1)

>> [B,A] = BUTTER(N,Wn)

– Designs an Nth order lowpass digital Butterworth filter
and returns the filter coefficients in length N+1 vectors
B (numerator) and A (denominator).
– The coefficients are listed in descending powers of z.
– The cutoff frequency Wn must be 0.0 < Wn < 1.0, with
1.0 corresponding to half the sampling rate.
– If Wn is a two-element vector, Wn = [W1 W2], BUTTER
returns an order 2N bandpass filter with passband W1 <
W < W2.

LAB 13: Noise Reduction 27

 Butterworth Filter Design
BUTTER (2)

>> [B,A] = BUTTER(N,Wn,ftype)

– Designs a lowpass, highpass, bandpass, or bandstop
Butterworth filter, depending on the value of ftype and
the number of elements of Wn. The resulting bandpass
and bandstop designs are of order 2N.
– [B,A] = BUTTER(N,Wn,'high') designs a highpass filter.
– [B,A] = BUTTER(N,Wn,'low') designs a lowpass filter.
– [B,A] = BUTTER(N,Wn,'stop') is a bandstop filter if Wn =
[W1 W2].

LAB 13: Noise Reduction 32

 Butterworth Filter Design
BUTTORD (1)

 Butterworth filter order selection

>> [N, Wn] = BUTTORD(Wp, Ws, Rp, Rs)
– N: the lowest order of the digital Butterworth filter
– Rp dB: maximum loss in the passband
– Rs dB: minimum attenuation in the stopband.
– Wp and Ws
• the passband and stopband edge frequencies
• normalized from 0 to 1 (where 1 corresponds to 𝜋
radians/sample).
• Lowpass: Wp = .1, Ws = .2
• Bandpass: Wp = [.2 .7], Ws = [.1 .8]

LAB 13: Noise Reduction 33

 Butterworth Filter Design
BUTTORD (2)
• Bandstop: Wp = [.1 .8], Ws = [.2 .7]
• Highpass: Wp = .2, Ws = .1

– Wn: the Butterworth natural frequency (or, the "3 dB

frequency") to use with BUTTER to achieve the
specifications.

LAB 13: Noise Reduction 34

 Butterworth Filter Design
Example : Lowpass Butterworth Filter (1)

For data sampled at 1000 Hz, design a lowpass filter with less
than 3 dB of ripple in the passband, defined from 0 to 40 Hz,
and at least 60 dB of attenuation in the stopband, defined
from 150 Hz to the Nyquist frequency (500 Hz).

>> Wp = 40/500; Ws = 150/500;

>> [n,Wn] = buttord(Wp,Ws,3,60); % Gives min. filter order
>> [b,a] = butter(n,Wn); % Butterworth filter design
>> freqz(b,a); % Plots the frequency response

LAB 13: Noise Reduction 35

 Butterworth Filter Design
Example : Lowpass Butterworth Filter (2)

LAB 13: Noise Reduction 36