how do i seperate two different frequencies in one sound?

I have an voice record. It contains a man's speech and another dominant voice with different frequency.

How can i seperate them? Also, how can i delete jammer effect of dominant tone voice.

Ex: https://imgur.com/a/mTRfek1

[KSSV]

Use fft get the frequencies.....and then use filters to remove the frequencies.

Read about fft, butterworth.

[Eqehan]

Ok, i am going to check fft and butterworth. Thx KSSV.

[Star Strider]

This could be a ‘blind source separation’ problem. The rica and ReconstructionICA functions could be appropriate.

[Eqehan]

@KSSV and @Star Strider thanks for your helps guys.

I completed the project. I used Fourier Transform(fft) and Band-Pass Filter. I am sharing the codes and screenshots for community.

clear all;close all;clc;                % Clear everything

[data, Fs] = audioread('sound1.wav');   % Define sound document
D= fft(data);                           % Fourier Transform of data of our sound
plot(real(D));                          % Plot the real part of fourier transform.

I found human voice frequencies at the result of fourier transform. It is between 1500-7500Hz. So i want to take just this frequencies. Then i made an Band-Pass Filter.

Band-Pass Filter

clear all;close all;clc;                % Clear everything

[data, Fs] = audioread('sound1.wav');   % Define sound document
N=128;                                  % Order of filter
Fn = Fs/2;                              % Nyquist Freq
fc1 = 1500/Fn;                          % Cut_Off Freq 1 - 1500Hz
fc2 = 7500/Fn;                          % Cut_Off Freq 2 - 7500Hz
B = fir1(N, [fc1 fc2], 'bandpass');     % Band-Pass Filter
ses_filtered = filtfilt(B, 1, data);    % Apply filter
sound(ses_filtered,Fs);                 % Play the filtered sound
audiowrite('sound_result.wav',ses_filtered,Fs); %record the new sound

So, the first subplot(2,1,1) shows my raw sound, which has noise in high frequency.

Second subplot(2,1,2) shows the result.

[Star Strider]

Congratulations!

[Eqehan]

Thank you mate.

Answer 1

[Eqehan]

I completed the project. I used Fourier Transform(fft) and Band-Pass Filter. I am sharing the codes and screenshots for community.

clear all;close all;clc;                % Clear everything

[data, Fs] = audioread('sound1.wav');   % Define sound document
D= fft(data);                           % Fourier Transform of data of our sound
plot(real(D));                          % Plot the real part of fourier transform.

I found human voice frequencies at the result of fourier transform. It is between 1500-7500Hz. So i want to take just this frequencies. Then i made an Band-Pass Filter.

Band-Pass Filter

clear all;close all;clc;                % Clear everything

[data, Fs] = audioread('sound1.wav');   % Define sound document
N=128;                                  % Order of filter
Fn = Fs/2;                              % Nyquist Freq
fc1 = 1500/Fn;                          % Cut_Off Freq 1 - 1500Hz
fc2 = 7500/Fn;                          % Cut_Off Freq 2 - 7500Hz
B = fir1(N, [fc1 fc2], 'bandpass');     % Band-Pass Filter
ses_filtered = filtfilt(B, 1, data);    % Apply filter
sound(ses_filtered,Fs);                 % Play the filtered sound
audiowrite('sound_result.wav',ses_filtered,Fs); %record the new sound

So, the first subplot(2,1,1) shows my raw sound, which has noise in high frequency.

Second subplot(2,1,2) shows the result.