Hi Spalding
The speed of change when using external sounds on any software instrument is determined on whether the sounds are stored in Ram or have to be loaded from the HDD, if they are stored in Ram then the changes are instantaneous, (Just as any voice on your PA1x) however if they have to be loaded off the HDD, then there is a delay (Dependent on speed of disc) while the sound is located and loaded into Ram.
Most VSTi these days allow pre-loading of sounds (The number of sounds being dependent on a combination of Ram size, CPU speed and HDD speed) into Ram so that changes can be instantaneous. (Third party hosts can also be used to pre-load sounds)
Note: Some older VSTi were designed purely for computer sequencers, and were not very effective when used live, however since about 2004/5 (When even cheap computers were really powerful) most VSTi have been optimised for live On-stage use, as well as for computer sequencers
Hope this helps

Bill