Sound Processing in MATLAB
What is digital sound data?
Getting
some pre-recorded sound files (digital sound data)
· You will first need the
following sound files. Right click on
these and choose “Save Link As…” and put these in a
folder and make sure you use the pull down menu option File àSet Path to put that folder in the MATLAB
path (or simply put the .wav files in your correct working directory.
flute.wav
Loading Sound files into MATLAB
· We want to read the digital
sound data from the .wav file into an array in our MATLAB workspace. We can then listen to it, plot it,
manipulate, etc. Use the following
command at the MATLAB prompt:
[road,fs]=wavread('road.wav'); % loads “the
long and winding road” clip
· The array road now contains
the stereo sound data and fs
is the sampling frequency. This data is sampled at the same rate as that
on a music CD (fs=44,100 samples/second).
· See the size of road: size(road)
· The left and right channel
signals are the two columns of the road array:
left=road(:,1);
right=road(:,2);
· Let’s plot the left data
versus time. Note that the plot will
look solid because there are so many data points and the screen resolution
can’t show them all. This picture shows
you where the signal is strong and weak over time.
time=(1/44100)*length(left);
t=linspace(0,time,length(left));
plot(t,left)
xlabel('time (sec)');
ylabel('relative signal
strength')
· Let’s plot a small portion
so you can see some details
time=(1/44100)*2000;
t=linspace(0,time,2000);
plot(t,left(1:2000))
xlabel('time (sec)');
ylabel('relative signal
strength')
· Let’s listen to the data
(plug in your headphones). Click on the
speaker icon in the lower right hand corner of your screen to adjust the
volume. Enter these commands below one
at a time. Wait
until the sound stops from one command before
you enter another sound command!
soundsc(left,fs) % plays left channel
as mono
soundsc(right,fs) % plays right
channel mono (sound nearly the same)
soundsc(road,fs) % plays stereo (ahhh…)
· Another audio format is the
.au file format. These files are read in
using
[lunch,fs2]=auread('lunch.au');
soundsc(lunch,fs2);
· To save an array as a .wav
file, use wavwrite( ). Use auwrite( )
for .au format output.
Let’s Mess With the
Signal
(perform
digital signal processing)
Reverse Playing
·
To
play the sound backwards, we simply reverse the order of the numbers in the
arrays. Let’s experiment with a small
array. Type in the following commands:
y=[1;2;3;4;5]
y2=flipud(y)
·
Note
that flipud stands for flip upside-down which
flips your array y and stores the inverted array in a new array
called y2.
·
Now
let's try it on one of our sound arrays:
left2=flipud(left);
soundsc(left2,fs)
Digital Delay Experiment 1:
·
Now
let's add an echo to our sound data by adding to each sound sample, the sample
from a previous time:
leftout=left; % set up a
new array, same size as old one
N=10000; % delay amount N/44100 seconds
for n=N+1:length(left)
leftout(n)=left(n)+left(n-N); %
approximately ¼ second echo
end
·
Note
that these arrays are large and it may take some time for the processing to be
completed. Compare the input and output
by typing the following after the cursor returns, indicating that the
processing is done:
soundsc(left,fs) % original
% Wait until the sound stops before moving to next sound
command
soundsc(leftout,fs) % signal with new echo
·
This
program first sets the output to be the input.
This is simply a quick way to initialize the output array to the proper
size (makes it operate faster). The loop
starts at n=10001 and goes up to the full length of
our array left.
The output is the sum of the input at sample time n plus the input at sample time n-10000 (10000 samples ago, 10000/44100 seconds ago since the
samples are spaced by 1/44100 seconds).
Try some different delay amounts.
·
Try
it in stereo and we will echo left-to-right and right-to-left!
out=road; % set up a
new array, same size as old one
N=10000; % delay amount N/44100 seconds
for n=N+1:length(road)
out(n,1)=road(n,1)+road(n-N,2); % echo
right-to-left!
out(n,2)=road(n,2)+road(n-N,1); % echo
left-to-right!
end
soundsc(road,fs) % original
soundsc(out,fs) % echo
Digital Delay Experiment 2:
·
Try
the following variation on this theme, which keeps adding to the signal itself
from 1000 samples ago slightly softened (multiplied by 0.8). Note that for this sound data the samples are
spaced by T=1/8192 sec (fs2=8192 samples/sec).
[lunch,fs2]=auread('lunch.au');
out=lunch; % set up a new array, same size as old one
N=1000; % delay amount N/8192 seconds
for n=N+1:length(lunch)
out(n)=.8*out(n-N)+lunch(n); % recursive
echo
end
soundsc(out,fs2)
% echo
·
This
echo process is like looking into a mirror and seeing a mirror with a
reflection of the first mirror, etc! The echo goes on forever, but gets slightly
quieter each time.
Digital Tone Control
·
The
following program (or “digital filter”) is designed to soften high frequency
components from the signal (treble). It
retains the low frequency components (bass).
Applying this digital filter has the same effect as turning down the
treble tone control on your stereo. The
design of this code is not so obvious.
The Electrical Engineering students will learn more about this type of
frequency selective digital filtering in ECE334 Discrete Signals and Systems.
[hootie,fs]=wavread('hootie.wav'); % loads Hootie
out=hootie;
for n=2:length(hootie)
out(n,1)=.9*out(n-1,1)+hootie(n,1); % left
out(n,2)=.9*out(n-1,2)+hootie(n,2); % right
end
·
Compare
the input and output as before. Note
that the modified signal sounds muffled in comparison to the input data. This is because the high frequency components
have been suppressed in the output.
soundsc(hootie,fs) %
original
soundsc(out,fs) % low pass filtered
·
A
small change in our digital filter allows us to boost high frequencies and
suppress low frequencies:
out=hootie;
for n=2:length(hootie)
out(n,1)=hootie(n,1)-hootie(n-1,1); % left
out(n,2)=hootie(n,2)-hootie(n-1,2); % right
end
soundsc(out,fs) %
high pass filtered
Changing the Speed
·
The
sampling frequency fs tells us how much time goes
between each sample (T=1/fs). If we play the song with more or less time
between samples than was originally there when recorded, the speed will seem
off, producing interesting effects...
soundsc(hootie,fs/1.5) % How slow can you go?
soundsc(hootie,fs*1.5) % The Chimpmonks!
Removing (Minimizing) Vocals
·
In
most popular music recordings, the vocal track is the same on the left and
right channels (or very similar). The volume of the various instruments are more unevenly
distributed between the two channels.
Since the voice is the same on both channels, what would happen if we
subtract one channel from the other and listen to the result?
soundsc(left,fs); % Original left channel
soundsc(left-right,fs); % Long and
winding road, virtually no vocal
Notice the voice is virtually eliminated…
·
Try
it with Hootie…
soundsc(hootie(:,1),fs); % Original left channel
soundsc(hootie(:,1)-hootie(:,2),fs); % Hootie, reduced vocal
·
You
still hear some vocal here because this song uses a stereo reverberation effect
and the echo moves back and forth between left and right channels (like our
stereo delay above). This makes the
voice unequal from left to right channels.