Our concept is that the MIR is kinda "glorified" mixing engine, with multiple input streams, taking care of positioning, room information, directivity, and sonical adaptions. All this will be unified under one intuitive graphical interface, which "knows" a lot about the VSL-instruments and -ensembles fed into it, without further user interaction. - Underneath, an "expert" layer will allow experienced and/or experimental users the access to more detailed sets of parameters, of course. - A single instance of this engine is what will be needed for the whole task.
The whole engine will be very demanding computation-wise, so no final decision has been made wether this will be a completely native or a DSP-based solution. Given these facts, I think it is quite obvious that we won't spread any rumours about its architecture before we know _exactly_ what it will look like. I hope that the fog clears soon, but don't expect a market-ready product before summer (2004, that is :-] ...)
Anyway - thanks for your interest.
/Dietz - Vienna Symphonic Library