Sure, if they vibrate at massively high frequencies (I'm not sure what the potential ceiling is on electrostatic oscillation) then they could potentially displace alot of air over time, but at the efficient end for the MEMS chip (seems to be 4khz for this application), these are high frequencies only, not that useful for humans choose to listen to, if you lower the frequencies, then the amplitude of each cycle, not just average displacement over cycles must increase according to basic physics (f = 1/T and f = c / λ). Anyone who's ever experienced low bass frequencies know this intuitively, you hear it, sure, but you also feel the high amplitudes in your belly. For low bass, there's massive volumes of air being moved per cycle. Can a visibly unmoving wafer move that kind of air? If the amplitude doesn't increase, but stays in the sub mm range, and you drive it with a standard audio signal, eg. speech/music, you'll only end up with higher frequencies with energy in them, not the lower frequencies, in other words you are distorting / modulating / modifying the sound wave and it's not a faithful reproduction, the sound is top heavy, like an old school telephone call. Just looking at the mathematics of theoretical possibilities in the video you sent, it's complex but I can see that it boils down to a stated loss of 6db per octave (octaves are at every double or halving of frequency), so starting from 51db at 4khz, by the time you get down to a middle C at ~262hz it's ~20db, which is like someone whispering from 5 feet away, and by the time you get to 20hz, the low end frequency speakers should reproduce, we are sitting on 1db (0 is the lowest sound pressure level possible for human hearing). Seems even if we can reach our theoretical limits, it's very unlikely we could produce practical audio applications replacing loudspeakers in similar or smaller footprints.
Unless we use possibility #6, but for that I need to do some more research.
Expand
(20min delay)