Laser microphones have actually been around for decades. Point a coherent light source at the outside of a window and the vibrations from inside will deflect that beam proportionally. Looks like the folks at MIT have come up with a truly novel way of capturing sound using only high speed video pointed at objects sympathetically vibrating in a sound field:
From discovery.com "Conversation Heard in Potato Chip Bag Vibrations"
I really loved that Shazam was able to determine the DNA of the song from one of these video sound captures! The audio interpolated from the 60FPS video really intrigues me, too -- I'm still trying to wrap my head around the physics.