Random vibration induces phase noise at low offsets from the carrier frequency, as shown by the difference between the blue (no vibration) and red (with vibration) curves in Figure 8. Although the ILSI MMD MEMS oscillator exhibits higher close-in phase noise when tested in a quiet environment, adding random vibration does not significantly increase the phase noise. In contrast, both SAW-based devices showed dramatic increase in phase noise under random vibration. This level of degradation can be detrimental to systems sensitive to close-in phase noise and shows how devices in real-world conditions may perform differently from datasheet specifications.
Yes, this looks impressive but if you look at the phase noise - the left hand side of the MEMs plot shows phase noise of -40dB @10Hz - a standard oscillator will have about -90dB@10Hz - a good oscillator about -120dB@10Hz. So before vibrational disturbances this MEMS (which is apparently one of the least susceptible to vibrational disturbances) is already 50dB worse than a standard clock & 80dB worse than a good clock - does ordinary vibration from music playback affect oscillators by introducing 50 to 80dB of noise - I doubt it?