mike_mccue
I don't have a dsp chip running in my ears... yet. 

Actually your ears do act as a dsp chip and they perform a process known as "lossy compression" before the signal reaches your brain.

You make a good point.
 
:-)

FWIW, I did go read, and try to understand what I can.
 
The second paper: http://www.cplire.ru/rus/informchaoslab/papers/iccsc04ak.pdf
 
Seems to be explicitly describing processing techniques in a radio receiver.

It's the same thing.
 
The basic idea is simple - the room is acting as a filter at the listening position. IOW the room is doing the exact same thing to the signal as DSP doing convolution with the room's IR.
 
To any extent that that isn't true, the criticisms are legitimate (such as any non-linear response isn't captured, a different listening position doesn't match the IR, etc).
 
But to the extent that it is, it's a purely mathematical problem and you are just trying to create a correcting filter that is the inverse of the filtering done by the room. If you can create a filter (the room's IR) that contains a very complicated response over time, can you or can't you create a filter with an inverse response to that? 

Thanks for chiming in Drew - saved me some typing while I was out to lunch!
 
Also, regarding radio transmission vs audio acoustics, both deal with waves propagating through a medium, and so *at that level* have very similar (if not identical) mathematics to describe them. The waves are different (transverse vs longitudinal) as are the propagation mediums and the mechanisms of distortion, but the lines of propagation and reflection still look the same when you model the propagation channel (e.g. as a filter).

"can you or can't you create a filter with an inverse response to that?"
 
I think that you can.
 
 
 
But, I'm not sure how I feel about the idea that a room will have a linear response with regards to any practical or real life experience. Isn't that why "anechoic" chambers were invented?
 
One would think that a rectangular room with bare walls can have predictable peaks and nulls right where the math says they will be, but every time I've run a RTA test in one I have found that the results vary from the anticipation.
 
Discounting absorption of wall surfaces and presuming constant surface reflectivity characteristics, I'd want to learn more about:
 
Is the impedance of air "linear"?
 
Is RT60 Frequency dependent?
 
How does varying *musical content* energize a room compared to an all pass sweep or chirp?
 
It seems to me that the synthesis, as a room is charged with energy featuring differing frequency components, will create varying interactions with results that differ from an idealized flat response.
 
 
 
That's why I keeping thinking, wrongly or rightly, that a system needs to have an active component to make corrections that achieve the results that are suggested by displaying a flat line.
 
If ARC et al. actually provided an opportunity to test its results instead of simply drawing a "predictive display line" I think it would be very informative. 
 
 
 
Corrections welcomed!!!

losguy
Thanks for chiming in Drew - saved me some typing while I was out to lunch!
 
Also, regarding radio transmission vs audio acoustics, both deal with waves propagating through a medium, and so *at that level* have very similar (if not identical) mathematics to describe them. The waves are different (transverse vs longitudinal) as are the propagation mediums and the mechanisms of distortion, but the lines of propagation and reflection still look the same when you model the propagation channel (e.g. as a filter).

I would think it would be even harder to predict the perfect FIR for radio response as the environmental circumstances are constantly changing.
 
Is that what the "decision-feedback equalization" schemes mentioned in the papers are helping with?

@Mike,
 
Anechoic chambers are not for reducing nonlinearity, they're for reducing, well, echoes as would come from waves reflecting off of surfaces. (Radio antennas are tested in anechoic chambers too, for the same reason BTW. Only the materials used there are to absorb radio waves.)
 
I took an interest in the idea you raised about of distortion due to the air medium itself. According to this ancient paper, it looks like you need about 100 dB SPL for things to get noticeable. (Of course, at that level, other things will be noticeable, too.)
http://hal.archives-ouvertes.fr/docs/00/21/95/66/PDF/ajp-jphyscol197940C860.pdf
They did the experiment at 20 kHz, but the effects should translate to lower frequencies, at least roughly.  I noticed that they had to take great pains to find a way of generating the acoustic waves at levels approaching 160 dB SPL without distortion from the transducer itself. Quite a feat, I'd say, even in this day and age (that is, unless someone just happens to have a 160 dB SPL acoustic sinewave source to tell me otherwise).

mike_mccue
I would think it would be even harder to predict the perfect FIR for radio response as the environmental circumstances are constantly changing.
 
Is that what the "decision-feedback equalization" schemes mentioned in the papers are helping with?

Yes, there are ways of casting the problem in an adaptive form, and at the receiving end only, too. There, you assume something about the target response that you're looking for, and "train" the system to always reach for that goal in the equalizer.  For communications systems, it can be about sharpening the shape of the "bits" as they come in. For speech, it can be about reducing echoes just over the speech bandwidth - and this is exactly what they do in the Polycom teleconferencing units and speakerphones (and all phones for that matter!).

mike_mccue
Discounting absorption of wall surfaces and presuming constant surface reflectivity characteristics, I'd want to learn more about:
 
Is the impedance of air "linear"?

 
http://www.sengpielaudio.com/calculator-air.htm
 

 
Is RT60 Frequency dependent?

 
Yes, but it is captured by the IR. 
 

How does varying *musical content* energize a room compared to an all pass sweep or chirp?
 

 
To get the IR, you want to make sure every frequency of interest is present and also want to minimize problems caused by ambient noise.
 
But you're right - no one wants to get up and dance to a sweep or chirp.