include loudspeaker array optimal order

[joslloand]

... characterization

[mtmccrea]

without looking closely, Archontis's HOA Matlab Library has a similar function, which may be of use (from Zotter and Frank's ALLRAD paper).

[joslloand]

Ah! That's useful to see. This function requires VBAP, which we don't have a current implementation of in sclang.

I'm under the impression that the latest version of the ADT uses a slightly different approach. (With a very quick browse, I'm not seeing it.)

My intuition suggests that a pseudo-inverse of the encoding matrix should be able to return similar / the same information. I was thinking that Heller was doing something along those lines.

If yes, we'll want to compare with the Zotter, Frank, Archontis results.

[joslloand]

Ah, here's another implemention, found in spaudiopy.

[mtmccrea]

spaudiopy was, at least in the beginning, largely a port of the HOA MATLAB library (though the author has gone on to add more things), and now tracks pretty closely with SAF. In this case it looks like the same implementation: the same operation on the average vbap spread throughout the collection of LS triplets. ADT may be smarter about directional variation, as Heller has done more with dome decoding, etc. (the huge LS gaps in a hemisphere would throw off the average spread), so it'll be good to check!

[mtmccrea]

This function requires VBAP, which we don't have a current implementation of in sclang.

I remember a while back I was working with VBAP code in SC, I think S. Wilson did a port of a max patch IIRC? There was a function that did the triangulating, but it wasn't bullet proof. I recall finding occasional errors when I plotted the results in 3D. That code is in SphericalDesign:-calcTriplets.
It would be interesting to see if any of these other libraries use the same or different triangulation algorithm, or if they're all original VBAP code...

[joslloand]

Ah... if SphericalDesign:-calcTriplets is what we need, that would be great!

[joslloand]

... another thought, intuitively, it would seem that just encoding loudspeaker incidences into SH should give some indication of overall distribution, just by observing what would be the resulting energy across degrees...