(Just thought I'd better bump this before the weekend!)

Basically you get the amount of energy within each band of frequency in the Bark scale which supposedly should match our perception.

Thanks fore your reply, Emmanuel.

I understand the bark scale, but I need clarification on the data your object actually outputs.

So, please tell me if I'm correct: The values in the list output by zsa.bark are the combined _magnitudes_ of the various fft bins that fall within a particular bark band. So to get a meaningful reading out of it should I convert the list output to decibels?

regards, T

The values in the list are combined energy (root of the amplitude) of all the bins that fall within each bark band indeed. You can convert that to dB if you want to have a non linear scale, but that depends on what you want to do.

Thanks for the reply--

do you mind having a quick look at this patch, which illustrates the average energy per band using pink noise?

It shows that (as one would expect) that there is a characteristic curve produced by the values of the bands (the response is not flat).

My question is, given that the amount of energy in each band is not equal, to have a more meaningful display (ie closer to human perception), should I compensate for this by dividing each list element by a normalizing factor, so that if pink noise is applied to it then the list will be flat?

Sorry about bumping this, don't mean to be pesky... just wondering if I'm "barking" up the wrong tree regarding the response of the bark scale to pink noise. If anyone can explain the significance of it, or the lack of, whichever the case may be.. T