On 2022-05-31, Carôt, Alexander - Prof. Dr. wrote:
(Before I answer or even post more questions on-list, I'll have to admit
I'm a long term, severe alcoholic. Trying to go into recovery, but much
in decline as of now. Take what I say with precisely that grain of salt
in mind. Shoulda admitted to it much earlier, true... :/ )
1.) I have a Sennheiser Ambeo Mic which gives me 4 channel Ambisonics
A-Format. In order to achieve 2-channel binaural Sound via Headphones
I have to use a respective binaural decoder (as part of e.g.
https://github.com/videolabs/libspatialaudio), however, is it correct
that I first have to convert it from A to B-Format ?
In general B-format is the interchange format between tools which is the
most convenient, and which different tools most understand. It's the
easiest to handle since it's so regular, mathematically speaking. So in
general, yes, go via B-format.
But since you mention boom mics — which are of much higher order and
more irregular than first order POA from a sounfield mic and more
irregular, and even binaural rendering as the eventual D-format — there
is a certain caveat here. You might be able to work directly from A to D
while retaining more fidelity on the way. You'd lose isotropy, but in
frontal work, you might gain also many more orders of directional
accuracy, locally, beyond what going via B (or C) admits.
That ain't then ambisonic. The ambisonic framework might help you
analyze what you're doing, but if you go from A directly to D, your
transmission chain is something besides.
2.) If 1.) applies what is the best way to convert it ? I found this:
http://www.matthiaskronlachner.com/wp-content/uploads/2013/01/2012-09-27-Kronlachner-homepage.pdf
and it states quite straight-forward equations to do it:
W = 0.5 * (A + B + C + D);
X = (A + C) - (B + D);
Y = (A + B - (C + D);
Z = (A + D) - (B + C);
Is this correct and is B-Format what is typically called the Ambisonics Sound
field ?
The general *idea* is about right, but the execution isn't perfect.
First, the first ordersound field mic is built for Gaussian quadrature
of four points, in 3D, using cardioids, ABCD. They are not in a
plane, so the amplification coefficient over the terms is not as simple
as you'd have it. You can scale the signals how you want, but the basic
scaling would have the .5 factor in all of the signals, WXYZ.
If you then want to end up with an average energy scaling, W will end up
being scaled down less than XYZ. These sorts of reasons are why even
B-format is ill-defined to date: what are you trying to scale *for*?
Finally, the really nasty thing here is the possibility of spatial
aliasing, particularly towards the higher frequencies. Because even with
the best classical SoundField mics, somewhere in the vicinity of
10-11kHz, the mic no longer physically works as a directionally coherent
soundfield mic. The wavelength of the sound being captured starts to
undercut the distance between the mic capsules.
That means that a truly excellent mic has to apodize in direction what
it measures, against a rising frequency. It has to do so both by
physical design, and the signal processing which leads from its A-format
to its B-format output. It cannot be a perfect B-format sensor, but
instead it has to compromise. Degrade gracefully at the higher
frequencies. And in order to do that in a controlled fashion, it cannot
just do a real matrix from A to B, but it needs to do a MIMO LTI matrix
of filters on the way.
That filtering can be done in the B-domain to a degree, as the early
SoundFields did it. But really, nowadays, it'd be better to do it via
DSP. To formulate the whole thing as an inverse MIMO problem.
--
Sampo Syreeni, aka decoy - [email protected], http://decoy.iki.fi/front
+358-40-3751464, 025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2
_______________________________________________
Sursound mailing list
[email protected]
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit
account or options, view archives and so on.
