Narrow band to third octave band conversion

Hi,

I'm trying to convert my narrow band data from narrow band frequency data into Third octave band data. I have two matrices, the first is a matrix of the third octave bands of interest; column 1 = lower frequency, column 2 = centre frequency and column 3 = upper frequency. The second matrix is the collected data; column 1 = narrow band frequency and column 2 is collected data.

Essentially, I'm looking to take the narrow band frequencies (column 1 of f_a) that is greater than the third octave band lower frequencies and less than upper frequencies, and take a mean average of the relevant data (column 2 of f_a), then assign that to the centre frequency in a seperate matrix.

For example, for the 100Hz third octave band, with the lower and upper frequencies being 89-112, take the mean average of f_a[68 : 84 ; 2] and assign that result to 100Hz. And all the third octave bands of interest to be placed in the same matrix if possible.

Any help would be appreciated.

[Keith]

Shame there is no answer to this query

Answer 1

[Mathieu NOE]

hello

I think I already answered a very similar question in the past

anyhow , this is one liitle function I wrote long time ago

this was for my own purpose , the fft spectrum was given in dB units

other wise , if your fft amplitudes are in linear units simply change this line

Slin_squared = 10.^(Sdb/10);	% conversion dB -> lin ^2

into

Slin_squared = Slin.^2;	% conversion lin -> lin ^2

I also found another one (attached) FYI

function [fto,Sto_dB] = narrowband2thirdoct(freq,Sdb)
%  fft narrow band to 1/3 octave bands conversion
% within each band , we do a energy sum (sum of the squared fft amplitudes)
% then the dB is obtained by 10*log10 of the sum
% M Noé
% 04/03/2010

Slin_squared = 10.^(Sdb/10);	% conversion dB -> lin ^2
%  fft narrow band to 1/3 octave bands conversion
fref = [10, 12.5 16 20, 25 31.5 40, 50 63 80, 100 125 160, 200 250 315, ...
    400 500 630, 800 1000 1250, 1600 2000 2500, 3150 4000 5000, ...
	6300 8000 10000, 12500 16000 20000 ];
	
ff = (1000).*((2^(1/3)).^[-20:13]); 	% Exact center freq. 	
a = sqrt(2^(1/3));	%
f_lower = ff./a;
f_upper = ff.*a;
ind1 = find (f_lower>min(freq));  ind1 = ind1(1); % indice du premier f_lower supérieur à min(freq)
ind2 = find (f_upper<max(freq));  ind2 = ind2(end); % indice du dernier f_upper inférieur à max(freq)
ind3 = (ind1:ind2+1);

for ci = 1:length(ind3)
    ind4 = find(freq>=f_lower(ind3(ci)) & freq<=f_upper(ind3(ci)));
    Sto_dB(ci) = 10*log10(sum(Slin_squared(ind4)));     % 1/3 octave band level in dB
    fto(ci) = fref(ind3(ci));                           % central frequency of the  1/3 octave band
    
end

[Keith]

Thankyou Mathieu!

[Mathieu NOE]

my pleasure !