2002-01-25 03:33:58 UTC
not so far been able to find concise answers to.
This may get a bit rambling, but I'll try to break the question(s) into
logically distinct parts. Anyone who replies please feel free to change
the subject line to represent the specific topic, since some of the
sub-questions are really completely separate issues.
1) Is anyone doing direct B format recording, meaning using two
figure-of-eight microphones in a M-S arrangement with an added omni
for the W signal?
I am told that some of the gorgeous Chesky recordings were done with
an AKG C-24 dual capsule microphone. I don't know if they were done
with capsule angles of 45/45, or 0/90deg., but either way I think that
anyone would be hard pressed to argue that you cannot get very good
sound with condenser figure-8 capsules. The C24 is discontiuued, but
AKG make a newer model with dual capsules.
Josephson even makes a microphone with two selectable
pattern capsules plus an omni capsule in the same housing, so the
technique could be used even without resorting to cumbersome multiple
microphone mounting arrangements like Nimbus seemed to use.
So why do all the ambisonic recordists seem to use the Soundfield?
2) What effect do non-coincident microphones have on ambisonic
The Soundfield microphone has the capsules as nearly coincident as
and the ambisonic papers I have been able to read seem to assume
capsules as the starting point for the mathematical derivations, yet Dr.
Moorer of Sonic Solutions and Jack Vad of the San Francisco Symphony
done experimental recordings which utilize what is basically the NOS
technique (hypercardioid microphones at 110deg. and 13cm to 17cm
separation) with an additional rear facing hypercardioid, from which
the horizontal ambisonic signals were derived.
The paper can be found at:
That paper is the first mention I have seen in print of non-coincident
ambisonic techniques (although admittedly I do not have a lot of
reference material which covers ambisonics in any detail).
3) Using the the technique described by Dr. Moorer, is any advantage
by using more than three microphones to generate the ambisonic signals?
I think that would in effect be similar to mathematically having three
unknowns but more than three equations constraining the results, i.e.
system is overspecified so you have some redundancy of information. In
practice it might have some benefit because you rely on the off-axis
performance of each microphone less, or it may have some disadvantage
you have more leakage between microphones, with the resultant phase
I am much too rusty on my matrix math to fully follow how Dr. Moorer's
equations could be extended to using four or five microphones. Does
have a reasonably educated guess as to whether the microphone technique
being developed by Jim Johnston at AT&T research labs is likely using
same idea? That array is described as 5 hypercardioid microphones, with
angular separation of 72 degrees between each, with two shotgun
pointed up and down. The array of 5 microphones separates the capsules
approximately "the diameter of a human head."
You can see a picture of JJ's array at:
-- Chris Caudle