How can I subtract two WAV audio FFT signals?

Hello MATLAB community,

Below, you will find 2 WAV files. One is the background noise and the other is the signal obtained during the experiment. I am interested in the 0-1000 Hz frequency range and as you can see in the below image, I figured out how to overlay 2 FFT's in 1 graph.

Now, I want to delete "noise FFT signal" from the "experimental FFT signal". Can I somehow subtract directly in the "Figure Window"? or I need to write an additional code?

In the example shown below, you can see a strong peak around 350 Hz, which I am looking at and by subtracting it from the background noise, I would get a smooth curve and obtain a raw experiment signal.

Finally, when I subtract 2 graphs, I could get just data points on the plot, which I want to join in a smooth line.

Thank you.

%clear, clc, close all

[y1,fs1]=audioread('Noise.wav');
t1=linspace(0,length(y1)/fs1,length(y1));
Nfft=8192;
% Nfft = length of fft
f1=linspace(0,fs1,Nfft);
X1=abs(fft(y1,Nfft));
%X = the fft of the samples y in 8192 points
plot(f1(1:Nfft/2),X1(1:Nfft/2))
xlim([0,1000])
 
%title('Combine Plots') 
%2nd graph begins

 hold on 
 
[y2,fs2]=audioread('Experiment.wav');
t2=linspace(0,length(y1)/fs2,length(y1));
Nfft=8192;
% Nfft = length of fft
f2=linspace(0,fs2,Nfft);
X2=abs(fft(y2,Nfft));
%X = the fft of the samples y in 8192 points
plot(f2(1:Nfft/2),X2(1:Nfft/2))

 xlabel('Frequency'); 
 ylabel ('Amplitude');
 title ('Background noise & experiment signals overlayed');

 xlim([0,1000])

hold off 

Answer 1

[Kiran Felix Robert]

Hi Avinash,

You can easily use a one line script to subtract the curves.

The data points in the plot functions are

plot1_x = f1(1:Nfft/2);
plot1_y = X1(1:Nfft/2);

plot2_x = f2(1:Nfft/2);
plot2_y = X2(1:Nfft/2);

If f1 and f2 are of the same length, then you can perform the following operation to subtract,

difference = plot1_y - plot2_y;

You can plot using the following code,

plot(difference,plot1_x)

If the vectors are not of equal length, you can always pad zeros to the end of the smaller length signal.

[Avinash Kandalam]

Thank you very much @Kiran Felix Robert. It works. :)