Originally Posted by

**Lupo**
The next subject in line that I

*believe* can be corrected with some certainty is the static reflections from speaker to boundary and back to speaker again. These alter the acoustic impedance as the driver sees it. They do not change with measurement position. It's also a straightforward process to calculate and experimentally verify the frequencies involved. As far as I've gathered, the boosts this cause can be corrected without much problem. Am I right in assuming this?

Quote:

Originally Posted by

**Brainchild**
I would also like to know more about this. The same issue that causes variable impedance at the speaker is simultaneously causing variable pressure around the room. If the speaker is putting out a linear response at the speaker, there still remains the issue of the nonlinearities at any given point away from the speaker.

Not a bad idea if one selects one position and focuses on one incident spot on a given boundary. But remember that reality is not quite so simple, and that we must address multiple sources and the incident point everywhere on every boundary - and suddenly the neat linear relationship becomes anything but. And the latency of subsequent processing becomes prohibitive even for just one listening spot and one source.

Now increase this scenario for all sources and increase the potential listening spots and the number of incident locations!

And lets not forget the complexities of reality that enter into the equation as all surface impedances are complex (this is a huge factor complicating issues) and that they are NOT constant and consistent.

Suddenly our ray tracing model is anything but a simple linear model. And we enter into the world where even the well defined room model (e.g. EASE and CATT-a) where auralization is employed is strained and which involves a latency that is utterly insufficient to support real time processing - even while assuming that the source is well behaved and functions as a true point source with defined Q and dispersion (well, we can dream, as this is ideal and "a consummation devoutly to be wished" - to coin a phrase!).

Hmmm...and does anyone dare to complicate the model by adding coupled spaces or any number of real world variations sure to make even a cluster of massively parallel processing systems shudder...

The fact is, we long ago left the realm of what is practically possible and entered into the realm occupied by room modeling that, for all of its amazing advances, is by no means close to real time processing and even less regarding the complex non-linearities of complex surface modeling and non-ideal sources.

The fact is, in practice we have difficulty doing this on the level equivalent to what an ideal model room mode calculator performs, let alone with the complex non-linear systems imposed by the variations of real world geometry and real world surface impedance models and real world source modeling applicable to a multi-positional time variant distributed topology that basically says 'do this extremely complex calculation once, and then an almost infinite more times for all of the various positional variations in near zero real time - assuming that there was a way to simultaneously modify the direct signal to address not only the direct signal, but to compensate for the

* multitude *of derivative reflected signals, modified in time and character by the non-linear nature of the incidence with a variety of complex surface impedances - and let's be nice and limit this model to simple first order reflections and ignore the complexities of such surface issues as

*diffusion*- of which 'simple' well behaved quadratic models are only now becoming available for simple modeling (e.g.EASE)!

Well...its easy to become overwhelmed in simply describing a very basic model, let alone in addressing the complexities of many real world environments. And we have already

*long ago* exceeded the capabilities of all but the most advanced modeling programs which do not have restrictions on latency (and which still exhibit fundamental simplification of surface and source models)!

And all of this assumes for the sake of discussion that the DSP algorithms are all fully mature, functional, and accurately address the myriad issues - as we have not even addressed the challenges of the mathematical models required to construct this Holodeck that many brochures are want to convince you not only exists, but exists in a mature form on readily available processing equipment that you simply run a jumper to or, better yet, exists in a few additional ICs mounted inconspicuously in the motherboard of your home theater receiver...despite the fact that we have strained or exceeded the capabilities offered by such systems as the RS/6000SP and derivatives running PSSP -assuming that you have the software models up and running!.

Bottomline, those wishing to see the basic capabilities of room modeling software easily costing more than most folks computing systems and processing capabilities increased (and based upon the input of carefully defined parameters - as opposed to supposedly measuring and determining their real nature on the fly with a 'built in' microphone!) are frustrated as much by how far things have progressed as by how far they have yet to go. And they have the luxury of allowing the programs to run for days on elaborate dedicated hardware platforms. And yet glossy brochures touting

*real time* "room correction" capabilities abound.

Dream as one might, and ignoring the fundamental kinks in the systems that form impediments to basic processing models, its time to acknowledge the fundamental limits to the systems.

And if anyone doubts this, start small and take some time and talk to the folks at AFMG or RPG regarding the difficulties they are having with basic room modeling models in both EASE and in programs as seemingly simple as Room Sizer an Room Optimizer - and ask them when the later two will allow input of parameters other than for simple ideal rectangular shapes, or heaven forbid, surface impedance - or to generate some real laughs, sources modeled as other than ideal point sources addressed in the far field!