Sound Pressure level returning -inf

Hello Community;

when I use the function [Lt,Leq,Lpeak,Lmax] = SPL(recording_vector) it returns vectors containing (partially or completely) -inf values. Surprisingly when I re-Run the same line I get usual numeric values. how is this possible? how to avoid it?

Thanks

[Chunru]

Can you show your data (a portion) or your code?

It is likely due to the signal amplitude to be 0 so log of it becomes -inf.

[Farid]

Thanks a Lot Chunru, I am attaching the file to this comment.

calib=importdata('calib.mat')
IRS=importdata('IRS.mat')

spl = splMeter("CalibrationFactor",calib, ...
               "FrequencyWeighting","Z-weighting", ...
               "TimeWeighting","Fast", ...
               "TimeInterval",0.125, ...
               "SampleRate",FS,...
               "Bandwidth","1 octave");

[Lt,Leq,Lpeak,Lmax] = spl(IRS)

Answer 1

[Jimmy Lapierre]

Leq, Lpeak and Lmax are only computed every TimeInterval, so they correspond to the SPL of zero for the first 0.125 seconds. And as chunru said, that corresponds to -Inf. When you call spl a second time, you are already pass the first 0.125 seconds of pure silence.

To avoid it, either discard the fist 0.125 seconds, or set a minimum value you prefer.

For example set any value below -100 dB (including -Inf) to -100:

Leq(Leq<-100)=-100;