Generate pure tone sequence in frequence domain

Question

D2snc 2021-10-24

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1570763-generate-pure-tone-sequence-in-frequence-domain

评论： D2snc 2021-10-26

Hi, i'm trying to generate a sequence of pure tones in the frequency domain and later on, convert to the time domain and execute the sound. The frequencies are 1000 Hz, 2000 Hz, 3000 Hz, 4000 Hz and 5000 Hz. I already managed to do this using only time domain, but i am having a very hard time to do this in the frequency domain. Can someone help me ?

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Bjorn Gustavsson 2021-10-25

Why not do it in the time-domain and Fourier-transform it.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Scott MacKenzie 2021-10-25

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1570763-generate-pure-tone-sequence-in-frequence-domain#answer_815748

编辑：Scott MacKenzie 2021-10-25

在 MATLAB Online 中打开

Seems you want to start by specifying your signals in the frequency domain, then convert to the time domain. I think this does the trick:

% system spec's (adjust as needed)

sRate = 16584; % sampling rate

sPeriod = 1 / sRate;

n1 = sRate/2; % number of samples in one-sided freq spectrum

% create spectrum in frequency domain

fOneSide = (1:n1)';

specOneSide = zeros(n1,1);

% add sinusoids with varying amplitudes

specOneSide(1000) = 1.0;

specOneSide(2000) = 0.5;

specOneSide(3000) = 0.4;

specOneSide(4000) = 0.6;

specOneSide(5000) = 0.3;

% plot the frequency spectrum

figure;

set(gcf, 'color', 'w', 'units', 'normalized', 'Position', [.2 .4 .6 .4]);

tiledlayout(1,3);

nexttile;

plot(fOneSide,specOneSide);

title('One-sided Frequency Spectrum');

xlabel('Frequency (Hz)');

ylabel('Magnitude Spectrum');

% create the two-sided frequency spectrum

specTwoSide = [flip(specOneSide(1:end)); specOneSide(2:end-1)];

n2 = length(specTwoSide);

f = (-sRate/2:sRate/n2:sRate/2-sRate/n2)';

% plot it

nexttile;

plot(f, specTwoSide);

title('Two-sided Frequency Spectrum');

xlabel('Frequency (Hz)');

ylabel('Magnitude Spectrum');

% ifft the two-sided frequency spectrum to get signals in the time domain

y = ifft(fftshift(specTwoSide));

t = (0:n2-1) / sRate;

% plot it

nexttile;

plot(y(1:100)); % part of the signal only (better visual)

title('Time Domain Signal');

xlabel('Time');

ylabel('Amplitude');

% play it

soundsc(y, sRate);

6 个评论
显示 4更早的评论隐藏 4更早的评论

Scott MacKenzie 2021-10-26

在 MATLAB Online 中打开

Yes, and as an example, if you add

n1 = 10*n1;

after the initialization of n1, you are increasing the size of the spectrum by a factor or 10. So, with this, each frequency is specified x10. To get 326.6 Hz, use

specOneSide(3266) = 1.0; % frequency = 326.6 Hz, amplitude = 1

D2snc 2021-10-26