# mixing two signals together

249 views (last 30 days)
Jorge Cantu on 28 Apr 2015
Commented: Image Analyst on 28 Jan 2022
Hi there, I am trying to mix two signals together in order to display the frequency spectrum. I am running into matrix dimension errors however. How do I fix this? I am fairly new to matlab so if someone could provide code that would be great! Here is what I have so far:
function pushbutton8_Callback(hObject, eventdata, handles)
% hObject handle to pushbutton8 (see GCBO)
% eventdata reserved - to be defined in a future version of MATLAB
% handles structure with handles and user data (see GUIDATA)
fs=8192; % sampling frequency in hz. The "laughter" samples were obtained using Fs=8192. You should not change this value here.
length(y) = 0:1/fs:70000;
lngth2=length(y); % get the number of samples
length(y) = 0:1/fs: 70000;
lngth1 = length(y);
lngth3=lngth1 + lngth2; % get the number of samples
t03 = lngth3/fs; % signal duration
t=0:1/fs:t03-1/fs; % create the time index (for t-axis), 0, ts, 2ts,....
ylau3=y(1:lngth3)'; % save the first lngth samples to a variable ylau.
% this step is optional
sound(ylau3,fs); % play the samples
fylau3=fft(ylau3,2*lngth3); % FFT of ylau, this results in the one-sided spectrum
fylaus3=fftshift(fylau3); % fftshift to make the spectrum plot symmetric about f=0, i.e., two-sided spectrum
df3 = fs/length(fylaus3); % create unit frequency step for the frequency-axis (horizontal axis)
fw3=[-fs/2:df3:fs/2-df3]; % frequency range is between -fs/2 to fs/2
y_db3 = 20*log10(abs(fylaus3)./max(abs(fylaus3)));%normalized magnitude spetrum in dB
% axis is in Hz
plot(fw3/1000,y_db3); % obtain the magnitude spectrum plot where frequency axis is in kHz
xlabel('frequency (KHz)');
ylabel('Magnitude (dB)');
title('Frequency spectrum');
grid;
##### 2 CommentsShowHide 1 older comment
Vittorio on 28 Apr 2015
When loading chirp or handel the variables y and Fs are loaded into the memory, so one does not want to overwrite them.

Vittorio on 28 Apr 2015
Edited: Vittorio on 28 Apr 2015
By the way, "mixing" in signal processing means multiplying. However after looking at the spectrum of the signals I am taking the guess that the problem is meant to show how signal spectra sum. To see what I mean try to run this, then see the difference replacing y3 = y1+y2 with y3 = y1.*y2
y1 = y;
y2 = y;
MinimumLength = min([length(y1), length(y2)]);
y1 = y1(1:MinimumLength);
y2 = y2(1:MinimumLength);
y3 = y1+y2;
NFFT = 2^nextpow2(MinimumLength);
f = Fs/2*linspace(0,1,NFFT/2+1);
Y1 = fft(y1,NFFT)/MinimumLength;
figure;
plot(f,2*abs(Y1(1:NFFT/2+1)))
title('Single-Sided Amplitude Spectrum of y(t)')
xlabel('Frequency (Hz)')
ylabel('|Y(f)|')
Y2 = fft(y2,NFFT)/MinimumLength;
figure;
plot(f,2*abs(Y2(1:NFFT/2+1)))
title('Single-Sided Amplitude Spectrum of y(t)')
xlabel('Frequency (Hz)')
ylabel('|Y(f)|')
Y3 = fft(y3,NFFT)/MinimumLength;
figure;
plot(f,2*abs(Y3(1:NFFT/2+1)))
title('Single-Sided Amplitude Spectrum of y(t)')
xlabel('Frequency (Hz)')
ylabel('|Y(f)|')

Image Analyst on 28 Apr 2015
Darned if I know. Does this help? If not, explain why not, then post an error message.

Vittorio on 28 Apr 2015
Edited: Vittorio on 28 Apr 2015
Your code as a number of very basic programming errors. For example:
length(y) = 0:1/fs: 70000;
This makes no sense, since length is a function that returns a value to you, while you are trying to create a vector and assign it to it instead. Are you trying to change the length of y?
If you are trying to mix two signals in a signal processing-type way you just need to multiply them:
y1 = y; % set chirp sound aside
y2 = y; % set handel sound aside
MinimumLength = min([length(y1), length(y2)]); % compare the two to find the shortest one
y3 = y1(1:MinimumLength).*y2(1:MinimumLength); % multiply them element-by-element, making sure they are the same length
Then you can do all the fft you want on y1, y2, or y3. To do that I suggest you read carefully through the documentation of fft, looking at the examples.

Søren Jensen on 28 Apr 2015
fist time answering a question, so i will just copy it down here :P
i did not read through the whole code, but i think i found an error:
length(y) = 0:1/fs:70000;
the command 0:1/fs:70000 basically gives you a vector with the initial value 0 that increments by 1/fs until it reaches 70000 (thus a vector of length fs*70000 = 573440001) and you are trying to put this into a single position in the vector "length".
i believe you are trying to do the following:
y = 0:1/fs:70000
length_y = length(y) % or size(y,2) if your pc also runs out of memory at this call
ALSO:
are you sure the signal is 70000 seconds long?? thats 70000/3600 = 19.44 hours!!
ALSO:
your code for plotting the frequency spectrum seems a little wierd.. if it doesnt work, take a look at this code:
Fs = 8192; L = length(your_signal); NFFT = 2^nextpow2(L);
f = Fs/2*linspace(0,1,NFFT/2+1); spectrum = fft(your_signal,NFFT)/L;
semilogx(f(1:your_choice),2*abs(spectrum(1:your_choice)))
you can also you plot() instead of semilog if you don't like the semilogarithmic plot
- hope this helps

shahryar ghayoor on 28 Jan 2022
Hello Guys.
I need to mix two audio signals and not adding them together. (want to have one to be the background of the other)
I know that both signals have different Amplitude and frequencies with different durations.
So, how am I suppose to mix them and make a single audio file that has all mixture of features?
I appreciate if you answer me.
Thanks
Image Analyst on 28 Jan 2022
then post your data files in a new question (not here). Also explain why adding and rescaling (averaging) does not produce the effect you want, and exactly what one signal being the background of another signal means (since you say it's not just the average of the two signals).