How to make narrow band noise

Hi, I'm trying to make a noise with a central frequency of 2000Hz and a range of 0.25 oct. The specific parameters are:

cf1 = 2000; % central fq

bw = 0.25; % bandwidth in octave

low_f1 = cf1 / 2 ^ (bw/2); % lower limit of the fq range

high_f1 = cf1 * 2 ^ (bw/2); % upper limit of the fq range

So how do I make a noise for 0.6s long in matlab that ranges from the lower to the upper frequency range with 2000Hz as the centre? Thank you!!

Answer 1

[Chunru]

cf1 = 2000;         % central fq
bw = 0.25;          % bandwidth in octave
low_f1 = cf1 / 2 ^ (bw/2);      % lower limit of the fq range
high_f1 = cf1 * 2 ^ (bw/2);     % upper limit of the fq range
fs = 8000;          % sampling frequency
% geberate a gaussian white signal
n = round(0.6*fs);  % number of samples
x = randn(n ,1);
% band pass the white noise to get the narrow band 
y = bandpass(x, [low_f1 high_f1],fs);
[p, f] = pwelch(y, 1024, 768, 1024, fs);
plot(f, 10*log10(p))

[Freya Guo]

Hi Chunru, thank you so much for your answer, but when I try to play the sound using

sound(f,fs)

, it doesn't sound like noise more like a tap or something. Could you explain why using those parameters for

pwelch(y, 1024, 768, 1024, fs)

Thanks again!

[Chunru]

To make sound:

sound(y,fs)

The following code is to produce the spectral estimation of signal y:

pwelch(y, 1024, 768, 1024, fs)

You can "doc pwelch" for more info. The first 1024 means to divide data into segments of 1024 samples. The segment will overlap with 768 samples. 1024-point fft will be carried out on each segment. Then the average power spectrum will be computed to give you the estimate of power spectrum for the random signal y.

[Freya Guo]

Thank you for your patience, that was really helpful! Have a nice day.

[Ankita More]

Hi I ran this code, I got the graph however stimuli could not be heard

[Chunru]

soundsc(y, fs)          % to normalize the volum