> I'm willing to evaluate the "color" given to recorded sound by
> different microphones and by the various types of baffles, windscreens, a=
nd
> mounting options=E2=88=91
An excellent and very useful project.
> I suppose the best way is to record a white noise,
> however I would have something really "white", not colored by a speaker. =
Or
> at least something with very wide and smooth spectrum, even falling like
> pink noise=E2=88=91.
>
> I'm curious to know if you have something to suggest for generating such
> kind of wideband noise without a speaker=E2=88=91.
I did a test of a home-made windscreen at a waterfall. Recordings with wind=
screen on and off, then compared the spectra. It's the difference you care =
about, not the flatness of the stimulus.
Proceeding from that, one can put pink (not white) noise on a speaker, and =
record with and without the windscreen or baffle. Then make analyses of eac=
h recording and subtract one curve (naked is the reference) from the other.=
It doesn't really matter that the speaker's output looks awful on the anal=
yzer (they all do).
Now that test will give you the windscreen attenuation or baffle coloration=
for one angle of incidence only. I think it's really in the diffuse enviro=
nmental noise that one hears the coloration of a mic rig.
In my lab I have a set of surround near monitors, and a set of distant "the=
ater" monitors. I have a Pro Tools session that I call up for mic calibrati=
on tests that puts uncorrelated* pink noise into all ten speakers at calibr=
ated levels so that each speaker contributes equal SPL at the measuring loc=
ation.
In the field I hear a definite wide-band boost of around 3 dB around 250 Hz=
with my Jecklin disk. This reminds me I should measure that.
-Dan
* Made by recording a pink noise generator for ten minutes, then slicing it=
up into a one-minute segment for each track. I loop the 10-track one-minut=
e playback.
|