Imagine that you’re making a video of an important speech or interview and your audio fails. You’re left with this beautiful picture, and no sound to go along with it (unless you’re running a backup recorder like you should!). New research from MIT has found that many everyday objects can actually act as recorders, and that sound can be extracted from them after the fact.
All this is done via high-speed video, where an algorithm analyzes the minute movements of an object then extracts the audio from that. These movements are so small that they’re imperceptible to the human eye.
While I’m sure the spooks from the CIA are all over this already, it’s not exactly ready for prime-time audio as you’ll hear in this TED presentation from Abe Davis. But the technology is simply fascinating. One particularly cool example had a researching singing “Mary Had A Little Lamb,” which was extracted from a bag of chips outside on the porch about 10 feet away (see the graphic on the left).
It also doesn’t stop with audio. The process allows someone to take a still picture and manipulate it in 3D in realistic directions, since it’s all based on the real-life movement of the object. This should be a boon to CGI pros and animators everywhere.
Well worth the watch!