I cannot think of any good way of capturing or playing back vector audio. Analog recording and playback could be seen as being vector based, but it's inherently lossy.
It's not like vector descriptions are not in use though - they're quite handy for generating deterministic signals. They don't take up much space and allow for output in arbitrary quality. Some lossless audio compression scheme probably also uses them (prediction + error).
I'd say that for audio for us mere mortals, 24/192 is plenty good enough.
24 bit samples allow for a dynamic range in excess of 140 dB. The best converters available today barely approach 130 dB, and that is with multiple channels (IIRC 4) in parallel. We're already at a point where things get limited by plain ol' thermal noise. A "measly" 100 dB of dynamic range already allows for live reproduction of a jack hammer (distance 1 m) at original volume while the noise is still at the threshold of audibility. Pop productions these days only use a tiny fraction of that anyway, and even more dynamic material rarely exceeds 20..30 dB, given that it's still supposed to be listened to.
192 kHz is already over 4 times the minimum required for reproduction of all the audible frequencies. It's still perfectly fine for those old CD-4 quadrophonic records. There are uses for yet higher sample rates, of course, but audio is not among them.
