My ears, which have been subject to both "aging [and] abuse by too much loud sound" (small airplanes, pistol range before hearing protectors became de rigeur), certainly can't tell the difference between sound which cuts off at 15 kHz (32kHz sampling) and 20 kHz (48 kHz sampling). Only the young (under 30?) can hear above 15 generally speaking. OTOH the drop from 16 bits to 12 represents a loss of 24 dB dynamic range (the difference between the noise floor and the level at which distortion sets in). Thus you are a)limited in the material you capture and b)must be careful about gain settings (you don't have as much margin for error). A good AGC (a contradiction in terms IMO) may help with the latter (and the former as well in some cases).