I read multi-window capability as meaning that you could choose your time window instead of using the hard-to-know-what's-actually-being-measured multi-time windowing. Darn.
The functionality that you are describing has been built into Smaart for nearly a decade (since v.7.0) - where you can have many measurements running each with a different FFT size (Time Window). The MTW is designed specifically to give you a better idea of what is going on - since you don't understand this I will try to shed a little light on it, and perhaps you can go and do some more research on your own. All professional audio analyzers I am aware of utilize some sort of decimated transfer function - for ex Meyer Sound's Sim, or Systune's TFC (Time Frequency Constant). Similar to Smaart, but not the same.
I recommend reading/skimming the Smaart user guide - perhaps there are other features that you didn't know existed:
All current Smaart v.7 documentation is located here:
http://www.rationalacoustics.com/support/800456-Smaart-v7-DocumentationWhen using a fixed FFT size, you are committing to one TC and resultant frequency resolution for the entire audio spectrum - this can be problematic when trying to determine signal to noise. Also remember we can better analyze reverberant to non reverberant energy in the case of Polar vs. Complex mag averaging - this is something that has been added in v.8 where 'Mag. Ave. Type' is no longer a global only setting - FFT size was never global in v.7.
Smaart's MTW utilizes multiple time constants from low to high to give a constant 48th oct. frequency resolution - resulting in ~840 frequency points across the audible spectrum, with 48th octave resolution from 60Hz up and 1 Hz resolution up to 140Hz. The part that is likely alluding to most is that the time constants are chosen to better distinguish signal to noise - to make the resulting trace easier to read, and more accurately display what is actually being measured in the situation. Now, remember frequency and wavelength are inversely proportional, so as we go lower in frequency we require a greater TC to accommodate for the greater periods we are dealing with, and vice versa in the high end. By better matching the TC to the periods of the energy they are accounting for, we have a better ability to window out late arriving signal - to determine what is signal and what is noise.
You can also see this in the Coherence trace, which when using MTW becomes a much more useful tool for detecting timing mismatches between the reference and measurement than any single-size FFT based measurement with comparable low frequency resolution. That being said, it becomes obvious that the coherence trace becomes your best measure of intelligibility when combining systems in real time.
I hope this helps. I've attached a screen shot comparing the various FFT sizes in Smaart using the same measurement pair. You can really see how the MTW response (top trace) shows a consistent freq. resolution as compared to any of the single-size FFT measurements. Note that Mag Smooth is off - these traces are the raw FFT data. Being that these are all near field measurements - what you can't see is the incredible penalty you take as you get out into the acoustic space. I invite you to set up this experiment on your own and note the excessive visual 'noise' that you see with a 32k FFT compared to 256 FFT @ 5kHz for example in an actual acoustical environment (6" from a studio monitor on my desk is hardly 'Real World', as you will find out).
One of the cool features that v.8's multi window capability does bring is the ability to view one window in Impulse Response Mode, and another in Real time, where as each window can only be one or the other.
