Thank you Joseph. 3+4 is a great workaround, I haven’t thought about shortening the regions and adding a very short quantization interval.
It works fine in this particular case, even though I can think of a few situations where the beginning of the phrase isn’t tight to the grid or doesn’t follow tempo at all. I guess one could set the quantization interval to time rather than tempo, but this task would quickly become very time consuming (no pun intended) in order to fine adjust each region.
Also, in such a case it would probably be recommendable to trim every clip in order to align the beginning of the region with the start of the waveform. This would inevitably make the process of aligning each clip on the grid in FMOD extremely difficult.
The easiest way to solve all the cases would be the one that had been suggested here (I just found it). And if not applied to an async instrument, maybe add a “play till end of region” option or something similar, to ensure that there are no interruptions even if the condition changes midway (or the event stops, there could be uses for that too).