That said, I'm not as sold on the non parallel surfaces concept as I used to be. Non parallel surfaces will not eliminate modes, it will make the modes less predictable. The existing models (that I know of) won't work on non parallel surfaces, which makes the room an expensive experiment.
I'd take a room that is roughly 23 x 16 x11 on a concrete slab with medium weight carpeting and go from there. Use of a distributed bass array and judicious use of the right kind of treatments in the right locations and you would have a really good room pretty quickly.