On 2012-01-27, Richard wrote:
Maybe it's just me, but I've read that before, and I don't read anywhere where it says HJ is UHJ.
You need to read the analysis which led to the UHJ specification. That "U" means "Universal". It's a rather wide decoding band within the Scheiber sphere which encloses most of the locuses of the three base standards, and also tries to be benign towards all of the nastier quadraphonic encoding standards of the time. The ping-pong-ones, you know...
Eventually UHJ encoding first took one, easy to implement analog locus within that band, and thanks to manual miscalculation and/or circuit design fault then had to be changed at least once before taking its eventual form. The one singular locus within the band that it is now. But it most certainly never left that band after it had first been implemented, and the band had been documented in full, even if I think it literally cut a corner of it after the early 45J work at some point.
From what I've read elsewhere, the phase angles of the front channels were tweaked to create 'HJ'
True.
Maybe I'm wrong and it's either a poorly written document, or I haven't a clue.
I'm very bad at history and my episodic memory is totally bunk, so take my comments with a large grain of salt as well.
Again, if it's a tweaked 'H' it had to be encoded. All I'm looking for is that equation.
This might be why there is a problem, here: originally there was *no* *single* encoding or decoding equation. There were just three (final, there were earlier prototypes as well) originating standards with very different encoding locuses, and wrt Ambisonics UHJ, first a general band within which encodings could work well, then an encoding in the middle of it which was held to be optimal, and then a pair of encoding/decoding locuseds which were close enough to the optimum to make the current UHJ locus, while being easy enough to be implemented in analog hardware. Thanks to hardware considerations, the locus was changed at least once, so that the current UHJ locus is contentious.
I seem to remember the change between the early (closer to ideal) and the later (cheaper to implement) UHJ locus mostly affects the upper band power oriented decoding, and there leads to somewhere between 1.4-3.0 degrees error in localisation. That having been deemed good enough even by perceptual testing in living room conditions.
I'll admit I've not attempted to decode it as UHJ, I tried that with 'H' ages ago, and that was awful
Personally, I've said before that I'd very much like to apply some statistical inference and heady signal processing to these signal sets. In order to determine whether the precise encoding is even discernible from all of the others. Then to automatically switch between different decoding algorithms, so as to derive the most perfect decode with any source material.
But truth be told, evenas I've never tried it, my back of the envelope calculation suggests you can't immediately detect the difference even with a statistical, matched detector which is much more sensitive than the human ear. So, in that regard, UHJ's compatibility decoding, in both its forms, ought to work pretty damn well for everything encoded in UHJ, and even for most of the stuff encoded against its three source standards.
(Over the entire length of a 3-4 minute single, using the proper Bayesian priors, and some Black, inverse-Volterra Voodoo, I'd bet you could separate each and every x-2-x system from each other, with x-n-x systems steadily becoming easier to distinguish as n grows. But I've never tried it, so this remains at the level of pure conjecture.)
-- Sampo Syreeni, aka decoy - [email protected], http://decoy.iki.fi/front +358-50-5756111, 025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2 _______________________________________________ Sursound mailing list [email protected] https://mail.music.vt.edu/mailman/listinfo/sursound
