Designing a home theater room for optimal acoustics We are building an addition on our home that will allow us to build a basement home theater room with some level of flexibility regarding its size and shape.  I've been reading somewhat conflicting information about room dimensions and shapes to provide optimal acoustics, and few sources seem to directly cite a scientific basis for guidelines.  I kind of get the feeling that this is more art than science.
My first question would be--how reliable is room modeling software (or manual calculations) in providing an accurate acoustic room response?  Simulation has the allure of helping optimize room shape, but only if it's a good predictor. 
I've read conflicting guidelines regarding whether or not angling a wall, or multiple walls might help minimize standing wave problems, and that they certainly complicate the analysis.  I've read in some places that room dimension formulas are valid and other places that they're not.  It seems pretty universally accepted that dimensions shouldn't be near multiples of each other.  Are there current best practices for room design that are universally accepted?  
For reference, the room interior width is constrained to be 11'11", length can be up to 23', and height can be as large as reasonably excavated in a home basement (6 to 10 ft).
 A: This is quite an old post but I believe it will be relevant for years to come, as this is a field that will "trouble" people as long as good audio reproduction is a wish.
To start with room modelling software. Nowadays room simulations are more accurate than when the question was posted, and they will always become better with passing years (the rate of improvement is another story though). One has to keep in mind though that the method of simulations has a major impact on the validity of the results.
In general, there are two big "families" of simulation techniques. Those are the wave-based and the geometrical. In the former, all (or at least most) wave phenomena are modelled so the results are very accurate, provided that the data used are accurate! This means that one has to acquire correct data for the surfaces and materials in the room and some of the needed values are not something published extensively by manufacturers (such as Young modulus, surface impedances, etc.). For more information of such methods, one can have a look at methods such as Finite Differences (FD) (see the paper Generalized Finite Difference Time Domain Method and Its Application to Acoustics by Wei, Wang, Hou and Dang), Boundary Elements Method (BEM) (see the book The Boundary Element in Acoustics by Stephen Kirkup) and Finite Element Methods (FEM) (most probably the paper The Finite Element Method in Acoustics can serve as a good starting point but I haven't laid my hands on it).
In the latter, the sound is treated as rays and the treatment is done from an energy point of view. This means that the wave nature of sound is neglected entirely and phenomena such as diffraction and refraction are not modelled (this is not always true though and there have been proposed methods to model diffraction such as the one described in An Efficient Auralization of Edge Diffraction paper by Lokki, Svensson and Savioja and Modeling Diffraction from an Edge Between Surfaces with Different Materials by Lokki and Pulkki). Nevertheless, it has been proposed that geometrical methods provide somewhat accurate results for high frequencies, where the interference patterns are very localised in the spatial domain well beyond our ability to discern differences (or that may make a big difference/issue). For more condensed information one can refer to the book Auralization - Fundamentals of Acoustics, Modelling, Simulation, Algorithms and Acoustic Virtual Reality by Vorländer.
Since most simulation methods compute the results with a more or less modified geometrical method, you have to be aware of its erroneous results in the "low-frequency regime". Now, what does "low-frequency regime" mean? This depends on the dimensions of the room. There are various proposals as to how to divide the spectrum into different regions but the most used out there (in my experience) is the Schroeder frequency, which is calculated by 
$$f_{s} >> 2000 \sqrt{\frac{T}{V}}$$
where T is the reverberation time and can be calculated by one of the available formulas (such as Sabine, Norris-Eyring or Fitzroy. For more information on this see Prediction of the Reverberation Time in Rectangular Rooms with Non-Uniformly Distributed Sound Absorption by Neubauer and Kostek) and V is the volume of the room. Below that frequency, it is better to treat the sound field as a wavefield (as opposed to an energy/ray field - excuse the slight abuse of term here) and use some wave-based method, or just ignore the results.
An alternative is to use a mode calculator such as the one proposed by Robert Harvey in their comment (or one from the huge plethora of such calculators out there). You have to keep in mind though that they most probably calculate the modes of the room without absorption being considered. The effect of absorption could potentially change the frequency the modes happen (the amount of change could also depend on the amount of absorption in the room) as well as the bandwidth of each mode (less absorption would produce sharp modes with distinct, well-defined frequencies). The shape most probably won't change but one may experience a bit more pressure build close to the boundaries of the room.
Now coming to the shape of the room. By tilting the boundaries/walls can help to make the shape of the modes irregular (compared to the rectangular room), but it will NOT make the modes vanish. The room will still exhibit modes in specific frequencies, but the shape of the modes, as well as the frequency each mode is happening, may change (the latter, most probably, only by a small amount). This is why you would still require to take into consideration the dimensions of the room. What one can achieve by dimensioning the room "correctly" (a magnanimous word used here!) is to distribute the modes as evenly as possible in the spectrum. This would result in an even energy distribution in the spectrum, which means an, as much as this is possible, flat frequency response. Still, the spatial distribution of the modes will not go away, and you would, most probably, still get some spatial irregularities in the frequency response.
To conclude this long "answer", I believe that using as many tools as you have available with careful consideration on their limitations is the way to achieve the best possible results. What I would do is to use some geometrical modelling tool to get a rough idea of the result for the "high-frequency regime", then use a different tool to calculate for the "low-frequency regime" (this could range from FEM software to a "simple" mode calculator) and try to combine them in the end. In addition to that, I would also consider some well known good practices that are available to everyone (like where to place absorbers and/or diffusers in the room after building it, loudspeaker placement for optimum reproduction, etc.) to improve the end result. Although this may sound trivial, it is not and requires a lot of iterations and testing of different designs.
In the end... I am not sure whether this is more art than science, but I believe it is more of an art to use science!
