But if you collecting acoustic data over a period of time, transient sounds (noise) average out, and the loud peak (gunshot)
Well yeah, if you want to pinpoint the source of a massive transient or the source of an annoying continuous whistling sound or a never-ending repeated playback of some secret message, this sensor could work. But it would not be very useful for recording private conversations.
But look at it from this side: a normal microphone measures pressure as a function of time, i.e. p(t). If their is only one source of sound, you can reconstruct the sound wave at the source. If there are two sources, it becomes impossible to distinguish. This transducer will generate three signals: vx(t), vy(t), and vz(t). If you have three sources of sound, then you could, with proper tuning and calibration, disentangle the signal and reconstruct the output of each of the sources individually. As soon as there are more than three sources, say in a pub with 20 conversations going on at the same time, you cannot do this anymore simply because there are more bits of data being transmitted than being transduced; you would need to solve a system of three equations with twenty unknowns.