Re: speaker layouts

You might be interested in some work I did on choosing the right
positions for loudspeakers to minimize the spatial aliasing (actually
spatial imaging is a better name in the case of rendering, keeping
with the DSP terminology).  With these designs you can use some sort
of least-squares + regularization technique for decoding without too
much trouble.  Spherical sampling above 1st order is a tricky problem
and very interesting that seemingly good layouts (such as Fliege or
minimum energy designs) are sometimes really poorly conditioned for
spherical harmonic analysis/synthesis.

I can provide MATLAB code (that requires the optimization toolbox) to
anyone interested.  The code also allows you to bias the array towards
better horizontal reproduction (say for a 1st order 3D, 3rd order 2D
design).  The paper is:

J. Atkins, “Optimal spatial sampling for spherical loudspeaker
arrays,” Acoustics Speech and Signal Processing (ICASSP), 2010 IEEE
International Conference on, pp. 97 – 100, 2010.
link: http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5496174&tag=1

Josh



2011/5/2 Jörn Nettingsmeier <[email protected]>:
> On 05/02/2011 11:50 AM, Marc Lavallée wrote:
>> Mon, 02 May 2011 06:59:40 +0200,
>> Jörn Nettingsmeier <[email protected]> wrote :
>>
>>> not native, but here's a very simple one that has been shoehorned
>>> into a third-order workstation:
>>> http://cec.concordia.ca/econtact/11_3/nettingsmeier_ambisonics.html
>>
>> I'd like to understand the "Versatile (and Quite Luxurious) Speaker
>> Setup" part of your article, and how it relates to your recommendation
>> on the number of speakers:
>
> first of all, this example was picked more or less out of thin air in
> order to show that there are really no limits on the type of layout, as
> long as it's more or less regular. it's no standard, and i must confess
> i have never mixed on exactly this setup (although i created a similar
> setup once, for a temporary installation).
>
>>> in production, use HOA. in reproduction, use as many speakers
>>> as you can afford, up to a limit determined by the order of the
>>> source material:
>>> 1st OA: no more than six
>>> 2nd OA: six or eight
>>> 3rd OA: eight, twelve if necessary
>>>
>>> it's now an established fact that too many speakers degrade
>>> the result for any given order.
>>
>> Is your "versatile" setup appropriate for 1st, 2nd and 3rd OA?
>
> if you can afford eight speakers on the horizontal plane, you get some
> kind of native 5.1 that is not totally wrong, certainly no worse than
> what you find in most homes (if you find 5.1 at all).
> and you can do native quad, you can do first order over four speakers,
> and first, second and third over all eight. lots of opportunities for
> a/b comparison.
>
> when i mix, i always run several decoders in parallel, with subsets of
> my maximum setup, to see how the mix degrades. so i work in third order
> (if i can - my home setup only does 2nd), but i will frequently fall
> back to first order on all speakers or just a subset, or even UHJ stereo
> - my goal is to find a middle ground that delivers good UHJ stereo to my
> customer and also retains ambisonic goodness for the future.
>
> so yes, i do care how it sounds in first order. but it's like a mono
> compatibility check, not like it's my primary target.
>
> almost all recordings i do, i think in surround and fold down to stereo,
> even if it is clear that the customer won't ever care for surround at
> all. only in cases of emergency will i use stereo-specific tweaks.
>
>> Does the elevated hexagon work with FOA, or is it only for HOA?
>> Is the octagon good for FOA, even if you don't seem to recommend it?
>> Finally: is it a good setup for domestic listening?
>
> this setup is very likely overdone for first-order. i haven't explored
> first-order periphonic much, but i swear by the regular hexagon for
> first-order horizontal (and it's also very nice for second order).
>
> in my experience, first-order horizontal over eight speakers is slightly
> but noticeably worse in the sweet spot than six, so not recommended for
> homes if you seldom if ever listen to third order material.
>
> for larger audiences, the game seems to be a bit different, although i
> don't quite understand why. i find 1st order over eight speakers covers
> a larger area more easily and uniformly than six, but the theory says it
> shouldn't, because outside the sweet spot, rV reconstruction is mostly
> not happening and everything relies on rE, which is degraded by using
> more speakers than necessary...
>
> if i were to build a periphonic rig at home (which i might, when i have
> the spare cash), it would be three on the floor, six at ear height,
> three on the ceiling, i.e. a slanted dodecahedron. should be good for
> second order periphonic, still small enough to be able to enjoy native
> soundfield recordings without ruining the rE too much, and no compromise
> for horizontal-only material.
>
> if you're not interested in sounds from below the equator (which means
> you'll miss all those exciting footsteps and running water sounds), you
> could try a hemisphere with eight and four and maybe an added zenith (or
> just six and three), but then you need some spherical harmonic alphageek
> to compute a decoder for you that doesn't pull upwards like crazy.
> i'm trying to learn how to do it, but i'm only beginning to understand
> the problem, nowhere near to getting it done myself.
>
> many people swear by stacked rings, and i can say that the one time i've
> had the chance to work on three rings-of-eight, i was amazed to find no
> noticeable phasing issues, even though the room was almost dead (rt60 of
> 0.3s down to at least 150hz, the lovely SPIRAL rig in huddersfield).
> stacked rings still feel a bit wasteful to me, and they have these holes
> at the bottom and up top (where the SPIRAL has recently added a zenith
> speaker), but the absence of phasing was quite striking. i have too
> little experience with rings to conjecture a general rule from this, but
> i'm looking forward to my next stacked ring setup...
>
> --
> Jörn Nettingsmeier
> Lortzingstr. 11, 45128 Essen, Tel. +49 177 7937487
>
> Meister für Veranstaltungstechnik (Bühne/Studio)
> Tonmeister (VDT)
>
> http://stackingdwarves.net
> _______________________________________________
> Sursound mailing list
> [email protected]
> https://mail.music.vt.edu/mailman/listinfo/sursound
>



-- 
Joshua Atkins
Ph.D. Candidate
Dept. Electrical Engineering
Johns Hopkins University
3400 North Charles Street
Baltimore, Maryland 21218
_______________________________________________
Sursound mailing list
[email protected]
https://mail.music.vt.edu/mailman/listinfo/sursound

Reply via email to