Henni,
The actual voice is loaded.
This approach also gives the user the choise to simply select a style and play with all the existing MIDI parts OR select a style and have that style LOAD any corresponing AUDIO/SOUND parts. Ofcourse with the pre-load, you have to wait a few seconds to play the style ... which is why I still suggest keeping the AUTOLOAD on at the begining where all relative AUDIO/SOUND parts are loaded so whenever you select a style, you can now INSTANTLY play it (as is the case with any other arranger) and not wait for any loading ...
However the only downside to this method is ... you have no way of knowing the % of memory left so if your RAM is already getting full (with all the AJAMSONIC voices already pre-loaded) and you now go to select a style which requires an extra voice whose size is larger than that left in RAM ....
We're working on another solution for this...but ideally the 1GB memory suggestion on the table seems best - where with the back up battery, you pre-load these voices only once and they are there each time the machine is powered on.
We will see :-)
AJ