On 2013-06-17, Jörn Nettingsmeier wrote:
deriving low-frequency, high-order information from tightly packed capsules (i.e. capsule distance small compared to wavelength) requires ridiculous amounts of gain.
Yes. I referred to that as the A-to-B noise amplification problem within the matrix.
at some point, you will only be amplifying random noise, without any meaningful results. so it is important to limit the lower boundary of a high order signal set.
Agreed. But how precisely does that affect the definition of the signal set so that it suddenly isn't at your given order?
Suppose that you'll puncture the perceptual noise floor with amplified mic noise at 3rd order at, say, 400Hz. Why can't you just shelf it down to 2nd order on the encoding side and/or even cut out some of the material altogether, to taste?
It's true that on the decoding side you have to worry about the precise order, in order to apply your psychoacoustics right. However, any combination of transfer signals ought to be equally valid, because that signal set is always referenced in fully physical units. It's just a description of yet another physical soundfield, at whatever order it happens to be, perhaps with fields of lower resolution mixed in. In all cases it's just a perfect description of a physical soundfield upto a given order, no matter what lower order, less well-directionally defined stuff happens within it.
So, the decoder might be hosed, true. But evenso, your encoder would be performing just as it should. What you encoded would be just right and it could be archived as-is, to await for a mixed order active decoder which could handle it Dandy. Not to mention it'd prolly sound plenty good over current decoders as well: it isn't as though we mind even second order directivity at the frequencies you are talking about; in a closed space zeroth order often does the job, and even in a fully open space, first order in-phase will typically do the job.
sure, for a separated eigenmike signal set, the necessary corrections can be made for arbitrarily low orders, but as soon as you have panned higher order sounds in your b-format master, it gets messy to the point that i don't want to go there...
I believe it only gets messy if you really need to do perceptual decoding. Otherwise, you can just do a systematic or an in-phase decode and be done with it. Those approaches then work the better the higher the order you're working at. Even at 2nd order and certainly at 4th order an in-phase decode will blow the socks off most pairwise panned formats, with so called "infinite panning accuracy". That's then for literally an infinite number of sources, within the soundscape, if you're so willing to sum them in. ;)
the same problem arises if you recklessly combine a POA microphone with higher-order spots (like i'm guilty of), but the spectral issues are a lot more manageable.
If they carry the same signal, that's reckless, because of the time issues. But if you just sink a coherent POA 4-vectr within a 2+ order signal set, I believe that is just fine: the result is a perfectly valid 2nd order signal set, which only decodes suboptimally because we expect so much of the decoder to begin with. Again, there's nothing wrong with the mixed signal set. It's only that the decoder capable of optimizing for directional sources just went from a passive matrix to an active one. No biggie, in this day and age, I'd say.
plus it's way easier to keep a four-channel first order signal set separate from the panned HOA set than a 32-ch fourth-order eigenmike set.
That's why we come down to just the "usual" 25ch 4th order ambisonic set, with no vestige of EigenMike or any other transducer present. We don't care about where the signal came from or where it's going, we only care about its semantics. Which are pretty clear as they stand: a time-variant Fourier-Bessel expansion of a given order of the air pressure field about a sweet spot, about the ambient average. No channels mentioned there. ;)
-- Sampo Syreeni, aka decoy - de...@iki.fi, http://decoy.iki.fi/front +358-50-5756111, 025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2 _______________________________________________ Sursound mailing list Sursound@music.vt.edu https://mail.music.vt.edu/mailman/listinfo/sursound
