There isn't going to be an abrupt cutoff between sound and not-sound, of course. But there'll be a range of pressures where it stops being as useful to think of the phenomenon as "sound" and starts being more useful to think of it as a vibrating thing occasionally imparting more/less energy to individual gas molecules.

My semi-educated guess is that's around when the mean free path of the gas molecules stops being small compared to the wavelength of the sound. But that's just a guess.