Scribd
Upload
16 views5 pages

Lab Filter Noise Music

lab filter noise music

Uploaded by

ashyam3
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
16 views5 pages

Lab Filter Noise Music

lab filter noise music

Uploaded by

ashyam3
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

% Application of a low pass filter

% to filter out noise in an audio file


% Author: Chao Wang
%
% Task: Complete and run the script.
% The places you need to add code start with a capitized word.
close all; clear all;

% "music_with_noise.wav" is a stereo audio file with two channels


% left and right, the audio file was from
% https://web.stanford.edu/class/me161/labs/filter_lab/
% download the file from Blackboard,
% make sure the file is in the same directory of this matlab script

Task 2

y(t) = x(t) * h(t)

Y(w) = X(w) * H(w)

|y(w)| =| x(w)| * |H(w)|

WRITE code to import the audio and plot its FFT as the 1st of three
subplots, see fft_star_wars.m for an example

[x,fs] = audioread('music_with_noise.wav')

x = 786432×2
0 0
0.0000 -0.0001
0.0001 -0.0000
-0.0000 0.0001
-0.0001 0.0001
-0.0000 0
0 -0.0001
0.0000 -0.0001
-0.0001 -0.0001
-0.0001 -0.0002

fs =
44100

% plot frequency content of an audio file


N = length(x);
X = fft(x); % generate N-point FFT (DFT spectrum)
% between the range of 0 and fs (sample frequency)
X = X/N; % divide by num of points to normalize FFT;
X1 = fftshift(abs(X)); % abs computes the magnitude of the spectrum
% fftshift will make the spectrum symmetric around
% zero frequency, i.e. shift from (0, fs) to

1
% (-fs/2, fs/2)
f = fs*linspace(0,1, N) - fs/2; % generate frequency vector (-fs/2, fs/2)
% for plotting

soundsc(x, fs) % sound is autoscaled as loud as possible without clipping


%pause % need any key to listen to next tone

figure
subplot(3, 1, 1)
plot(f,X1);ylabel('|X(f)|'); xlabel('Frequency (Hz)')
title('Frequency Content of Input Signal')
xlim([-600 600])

build filter
The code below shows an example on how to build and use a digital filter. CHANGE the cutoff frequency below
to properly filter the noise

cutoff = 500; % cutoff frequency (Hz) for the filter

% build a 5th-order butterworth low pass digital filter


% to keep the music but filter out the noise
% b: numerator coeff of system transfer function, a: denominator coeff
% system transfer function is discussed in section 9.7 of the textbook.

2
%
% In the command below, 5 is the order of the filter, the second parameter
% specify the cutoff frequency (with value between 0 and 1), which is
% normalized to half the sample rate, 'low' denotes the type of the filter,
% i.e., low pass filter

[b, a] = butter(5, cutoff/(fs/2), 'low');

sys = tf(b, a, 1/fs); % filter system transfer function

H = freqz(b, a, f, fs); % get frequency response of filter

subplot(3, 1, 2)
plot(f, abs(H)) % plot frequency response (magnitude) of filter
ylabel('|H(f)|'); xlabel('Frequency (Hz)')
title('Filter Characteristics')

Signal processing: pass the input through the lowpass filter


The following three methods should give the same output. You can try each one by commmenting out the other
two Method 1 (most commonly used)

y = filter(b, a, x); % using difference equation

% % Method 2: y[n]=x[n]*h[n]

3
% h = impz(b,a); % get impulse response
% y(:,1) = conv(x(:,1),h); % time domain convolution
% y(:,2) = conv(x(:,2),h);

% % Method 3: Y(e^jw)=X(e^jw)H(e^jw)
% f = 0 : fs/(N-1) : fs;
% H = freqz(b, a, f, fs); % freq response
% Y(:,1) = X(:,1).*H'; % freq domain multiplication
% Y(:,2) = X(:,2).*H';
% y = real(ifft(Y)); % inverse Fourier transform

soundsc(y , fs) % sound is autoscaled as loud as possible without clipping

plot frequency content of filter output


N = length(y);
Y = fft(y)/N; % generate FFT and divide by num of points to normalize FFT;
f = fs*linspace(0,1, N) - fs/2; % generate frequency vector
Y1 = fftshift(abs(Y)); % magnitude of the double-sided spectrum

subplot(3, 1, 3)
plot(f,Y1);
% uncomment the following line to zoom in to a particular freq range
% xlim([0 1000])
ylabel('|Y(f)|'); xlabel('Frequency (Hz)')
title('Frequency Content of Filtered Output Signal')

4
5

You might also like