Asked by John
on 6 Oct 2012

Hi everyone,

I am using a simple code to extract the PSD from a set of time-based sound data. I am now interested in obtaining the octave band averages of this data and I am completely lost. Can anyone help me out?

The code I ave at the moment is:

*clear all

close all

echo on

Fs= 4000; % Sampling Frequency (Hz)

x = xlsread('No Excitation.csv',1,'J35:J52034');

Hs=spectrum.welch;

psd(Hs,x,'Fs',Fs);

echo off;*

Thanks, John

Answer by Wayne King
on 6 Oct 2012

Edited by Wayne King
on 6 Oct 2012

You want to average the power over octave bands? You can do that with the avgpower() method.

t = 0:0.001:1-0.001 Fs = 1000; x = cos(2*pi*50*t)+1/2*sin(2*pi*200*t)+randn(size(t)); psdest = psd(spectrum.welch,x,'Fs',Fs);

Now to get the average power from [0,100] Hz

pwr = avgpower(psdest,[0 100]);

It may be more useful to consider the percentage of the total power in that interval

totpower = avgpower(psdest,[0 Fs/2]); 100*(pwr/totpower)

John
on 8 Oct 2012

Thankyou. That is extremely useful. I do have one more question though if you don't mind.

I am using msspectrum to plot the sound power over the frequency range and using avgpower on my calibration tests to determine how much each test needs to be adjusted.

ie. adjustment = real calibration value - avgpower(calibration)

At this stage, how do I take that adjustment value and apply it to my tests.

ie. instead of plotting by hmss = msspectrum(Hs,x,'Fs',Fs);

plot(hmss)

I want to plot that exact plot by translated down by the adjustment value.

Also, is there any way of exporting the (x,y) values that build the graph if i do plot(hmss)?

Opportunities for recent engineering grads.

## 0 Comments