Scribd
Upload
48 views4 pages

Fast Fourier Transform in MATLAB: Magnitude of The Complex Amplitude

This document discusses how to analyze the frequency components of a sound using the Fast Fourier Transform (FFT) in MATLAB. It explains that sounds are made up of multiple frequencies that can each have different amplitudes. The FFT converts a sound from the time domain to the frequency domain, showing the contribution of each frequency component. In MATLAB, fft() returns complex amplitudes, and taking the absolute value and dividing by the length gives the magnitude of each component. Plotting these magnitudes against their frequencies from 0 to the Nyquist frequency reveals which components are strongest. The code finds the maximum magnitude and its frequency, the "principle frequency", to indicate what note is loudest.

Uploaded by

Samuel Taye
Copyright
© Attribution Non-Commercial (BY-NC)
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
48 views4 pages

Fast Fourier Transform in MATLAB: Magnitude of The Complex Amplitude

This document discusses how to analyze the frequency components of a sound using the Fast Fourier Transform (FFT) in MATLAB. It explains that sounds are made up of multiple frequencies that can each have different amplitudes. The FFT converts a sound from the time domain to the frequency domain, showing the contribution of each frequency component. In MATLAB, fft() returns complex amplitudes, and taking the absolute value and dividing by the length gives the magnitude of each component. Plotting these magnitudes against their frequencies from 0 to the Nyquist frequency reveals which components are strongest. The code finds the maximum magnitude and its frequency, the "principle frequency", to indicate what note is loudest.

Uploaded by

Samuel Taye
Copyright
© Attribution Non-Commercial (BY-NC)
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/ 4

Fast Fourier Transform in MATLAB

I
n sound, frequency (pitch) and amplitude (related to volume) are the principle elements
perceived by humans—both of which usually vary with time. Most sounds are composed of
more than one frequency at a time. In music, this is known as a chord. If this were not the
case, then we would be left with those annoying monophonic ringtones from the first cell phones.
You know, the ones where there were only 10 ringtones from which to choose and the odds of
more than one person having that annoying beeping rendition of Für Elise were very high. But I
digress.

In MATLAB, if we plot the amplitude data vector against time we get a plot similar to the one
below.

[data fs] = wavread('sweetchild2.wav');


t = linspace(0,length(data)/fs,length(data));
plot(t,data)
xlabel('Time (seconds)')
ylabel('Normalized Amplitude (-1 to 1)')
title('Time Domain Plot of Sound')

Time domain plot of the intro to “Sweet Child of Mine.”

This plot shows us how long the sound is (2 seconds) and how the amplitude (effectively related
to volume) varies over time. But it also can be very useful to know how each frequency
component of the sound contributes to the overall amplitude. For example, in most music the
melody line has larger amplitude than the other frequency components of the sound. Think of it
as if 3 different instruments played 3 different notes (frequencies) each at a different volume. We
can distinguish which instrument was playing which frequency at what volume using MATLAB.
This can be done by doing a Fourier Analysis of the sound data—a transform from the time
domain to the frequency domain.

Magnitude of the Complex Amplitude


In MATLAB, there is a built in function, fft( ), which can be used to analyze the frequency
components of a sound. More specifically, for each frequency we can measure two values:
amplitude and phase. Phase is a more complicated topic covered in other courses. It is very
common for these two values be represent together as one complex value called a complex
amplitude or phasor. This representation is derived from Euler’s Relation (not very relevant to this
course, but for arguments sake it is shown below).
1 1
A cos 2𝜋𝑓 = 2 𝐴𝑒 𝑖2𝜋𝑓 + 2 𝐴𝑒 −𝑖2𝜋𝑓

Don’t worry if you don’t understand this. It is not too important.


Because it is a complex number, it has both real and imaginary parts. For this course, we generally
are only concerned with the overall magnitude of the complex amplitude. Note that the real and
imaginary parts are not amplitude and phase, respectively. Therefore, it is incorrect to use the
real( ) function instead of abs( ) to achieve the magnitude of the complex amplitude.
Additionally, the fft() function returns the complex amplitudes scaled by the overall length of
the data. Thus, we’ll need to divide by the length of the data. Finally, we note from the above
equation that we need to multiply the magnitude by a factor of 2 to achieve the original
amplitude, A. All of this combines to give us the following line of code.

y=2*abs(fft(data))/length(data);

From this code, y will be a vector of the magnitudes of each frequency. However, the second half
of this vector contains data not relevant to the course. So the next line we run is:

y = y(1:end/2);

Nyquist Frequency
The frequencies that correspond to these magnitudes can be made by the following line of code
(where Fs is the sampling frequency of the sound).

x=linspace(0,Fs/2,length(y));

These frequencies vary from hertz to Fs/2 Hz. Fs/2 is also known as the Nyquist Frequency—or
the maximum frequency that can be perceived in the sound at a sampling frequency of Fs. In
other words, if you try to record a sound with a frequency of 10 kHz in it at a sampling frequency
of 15 kHz, the sound will not be accurately played back, according to Nyquist’s Theorem.

We are now ready to plot frequency versus magnitude:

plot(x,y)

Principle (Strongest) Frequency


It is often useful to know what the strongest (loudest) frequency in a sound is. In our example of
the three musicians playing 3 different notes this would indicate which musician was playing the
loudest. This can be done by taking the max( ) of the y vector which gives us the value of the
strongest magnitude and the index at which it occurs. We can then look up the frequency that
goes along with that magnitude by indexing the x vector with that index:

[y_max index] = max(y);


f_principle = x(index);

We can modify our plot command above to include a circle around the Principle Frequency:

plot(x,y,f_principle,y_max,’o’)
Summary
All of the above can be difficult to understand. Some of it is even beyond the scope of our course.
However, it provides a basis for understanding frequency domain analysis of sounds in MATLAB.
Below is an outline of what you should “take home” about the FFT:

 Sounds often contain more than one frequency at once.


 Each frequency component can contribute a different magnitude to the overall amplitude
of the sound.
 Time domain plots say nothing about the frequency components of a sound. A Fourier
transform is needed to do frequency domain analysis. The fft( ) function in MATLAB
does this for us.
 The output of the fft( ) function by itself is a vector of complex numbers.
 y=2*abs(fft(data))/length(data);
y = y(1:end/2);
o returns a vector of the magnitudes of each of the frequencies’ contributions to the
sound’s amplitude (for the frequencies we care about in this course)
 The highest frequency that can be perceived in a sound is given by the Nyquist Frequency:
f_nyquist = Fs/2;
 The frequencies that correspond to the y vector range from Hz to the Nyquist
Frequency and can be generated by:
x = linspace(0,f_nyquist,length(y));
 The strongest (loudest) frequency (f_principle) and its corresponding magnitude
(y_max) can be determined by the following:
[y_max index] = max(y);
f_principle = x(index);
 Frequency versus magnitude can be plotted (along with the Principle Frequency circled)
by:
plot(x,y,f_principle,y_max,’o’)
 The following code will make 2 plots: time versus total amplitude and frequency versus
frequency component magnitude.
Completed Code

%Read in the sound


[data Fs] = wavread('sweetchild2.wav');
%Generate a time vector that corresponds with data
t = linspace(0,length(data)/Fs,length(data));
%Plot time versus amplitude data
plot(t,data)
xlabel('Time (seconds)')
ylabel('Normalized Amplitude (-1 to 1)')
title('Time Domain Plot of Sound')

%Analyze the frequency components.


y=2*abs(fft(data))/length(data);
%Remove non-relevant data.
y = y(1:end/2);
%Calculate the maximum frequency that can be perceived.
f_nyquist = Fs/2;
%Generate the frequency vector that corresponds with y
x=linspace(0,f_nyquist,length(y));
%Find strongest magnitude
[y_max index] = max(y);
%Look up the frequency that corresponds with the strongest magnitude
f_principle = x(index);
%Plot frequency versus frequency componant magnitude, circle f_max
figure
plot(x,y,f_principle,y_max,'o')
xlabel('Frequency (Hz)')
ylabel('Normalized Magnitude (0-1)')
title('Frequency Domain Plot of Sound')

Note: the sampling frequency for this sound was 44,100 Hz. The Principle Frequency was about
415Hz which is approximately a “G#” note, the loudest note played in this recording of the intro to
“Sweet Child of Mine” by Guns N’ Roses.

You might also like