Raimund wrote:
> I just found the time to look at your file. Certainly the general documen=
tation at the beginning of the file is quite useful for beginners.
>
> I would only disagree a bit with the 5 dB correction between A-weighted a=
nd broad-band noise figures (you obviously took that from the RaneNote arti=
cle). In my experience, the difference between broad-band (20 Hz - 20 kHz b=
andwidth) and A-weigthed noise figures is only about 2 to 3 dB as long as t=
he frequency spectrum of the noise floor is relatively flat.
Yes, I used 5dB as that seemed to be the worst case figure I found discusse=
d... although as we know certain cyclical noise patterns could be worse yet=
! In other words, there is no simple worst case. :-)
I acknowledge that it is impossible to pick one figure to convert from A-we=
ighting to how we hear noise. It is a relation across frequency after all.
Feel free to change the value in the code form 5dB. If there is some consen=
sus that it should be less as a general rule, I too will change it. Otherwi=
se I think leaving it at a larger value will result in results with a highe=
r degree of confidence, which takes into account the presence of some degre=
e of noise more offensive to us than broadband noise.
-- Robin Parmar
|