When multiple conversations happen at once, the voice recorder can't recapture any individual conversations at all. Is there a way to separate out the two streams? Or are there any digital voice recorders that do it automatically?
While it's possible to do a little bit of this with clever EQ, separating multiple conversations out from each other on the same recording is a non-trivial task: You could spend a lot of time doing it, and your results would still not be very good.
Some human voices may be higher than others, but they all coexist in more or less the same range. If you had two voices that were vastly different on the same recording, you might be able to do a decent job of making them clearer. But several people all speaking at once can't easily be separated out, or not with ant sort of true separation.
There are no recorders that, to my knowledge, do this automatically. You'd need to capture each conversation separately on a multitrack recorder to truly have this capability. (Getting one may be within your budget; any modern laptop can handle this, running free multitrack software and an inexpensive USB interface, with a few mics plugged into it.) A multitrack recorder is any recorder that captures several separate streams of audio on parallel tracks of tape (or on a digital file) and plays all the tracks back in sync.