On 02/22/2016 11:24 AM, Peter Lennox wrote:
Following on from discussions of decoder solutions: Forgive me if
I've missed this (I've been watching sursound for about 20 years, or
so - but I just may have missed the odd discussion!)
Has anyone systematically studied the interactions between decoders,
speaker layouts and particular rooms?
I ask because, it seems to me that interactions between room
acoustics and speaker positioning are known to have significant
psychoacoustic (and aesthetic) effects. Informally, I've observed
ambisonics working better than it had any right to, in particularly
difficult rooms (big reflective empty shoebox, for example).
But in respect of particular speaker layouts, (as per the discussion
on avoiding too many speaker in the horizontal plane), it seems to me
that there could be non-trivial interactions, so that (for arguments'
sake) a particular room might benefit from 'this' speaker layout as
against 'that' speaker layout.
I would seem a monster task to test a wide variety of rooms each with
a wide variety of speaker layouts (and I haven't even mentioned the
possible variety of speaker dispersion characteristics!) - but in the
long run, it needs doing - and sufficient testing might reveal
'families' of layout-room acoustic relationships that can point to
underlying causal rules.
If it's been done, I'd like to read it, and if it hasn't - sounds
like I've just knocked together a precis proposal for a PhD project!
to me, the single most obnoxious effect is the phasiness. next is
localisation precision, but a long way down. absolutely nobody cares
about localisation accuracy, so optimising for minmum angular error
seems undesirable to me (unless there is a very specific use case).
i've been thinking about modelling the interference patterns to minimize
phasiness in a systematic way. haven't done that yet, but now that I
learned how to work with the SFS toolkit, i will look at it.
after setting up many ambi rigs, these are my working hypotheses:
* don't be precise. measure to centimetres, but then add random delay
errors in for subjectively nicer reproduction at low orders.
this is what needs to be tackled systematically. we need a rigorous
technique to dither spatial aliasing optimally.
i guess we want the peaks and dips to be smeared out uniformly over a
large area.
* reverberation is your friend. it smoothes away the phasiness. unless
your content has subtle reverb which would be drowned in the room response.
* 3D rigs seem to localise worse but sound better for low orders, my
guess is it's because of smoother interference patterns. their path
length variation is different to that of horizontal speakers across the
listening area, which might be helpful. that's why i tend to recommend
3d rigs over higher-order horizontal-only ones, unless you're sure that
all your content will be at maximum order. even then, some height is
nice. at ICSA 2011, i heard a small IOSONO wfs rig being augmented
rather haphazardly with just four small height speakers (which, to the
best of my knowledge, were used "to taste" and not in any systematic
way), and the improvement was absolutely striking, not for localisation,
but for tone color and plausibility of space.
* i have a hunch that stacked rings, for all their wastefulness, seem to
have very nice interference patterns. for example, the SPIRAL in
huddersfield (triple octogon with zenith) is in a ridiculously dry room,
but i was quite surprised about its first-order performance, especially
since it has waaaaay to many speakers for that, in theory.
if somebody can suggest any measurements that are feasible in the field,
i will gladly obtain them from any future rig that i get to set up. i
guess room impulse responses would be the most important piece of the
puzzle. maybe we should just sweep each speaker into a tetramic in the
sweet spot as a start. with careful analysis, that should contain a lot
of information about the speakers and the room. we get free-field
response above a few hundred hertz, and below the schroeder frequency,
we're in mode land anyways...
--
Jörn Nettingsmeier
Lortzingstr. 11, 45128 Essen, Tel. +49 177 7937487
Meister für Veranstaltungstechnik (Bühne/Studio)
Tonmeister VDT
http://stackingdwarves.net
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit
account or options, view archives and so on.
