Researchers learn how to eavesdrop through soundproof glass using a potato chip bag and plant leaves

The latest breakthrough in surveillance technology proves that it’s possible to eavesdrop on a target behind soundproof glass using a mere potato chip bag, household plant, or anything at all. Wait, what? Researchers from MIT, Microsoft, and Adobe have developed an advanced computer algorithm capable of analyzing the minuscule vibrations on an object’s surface and recreating a complete audio signal. In the first series of experiments, the team successfully recreated intelligible speech from the vibrations in a potato chip bag placed as far as 15 feet away.

How is this possible?

Recall that sound is a mechanical vibration that travels through matter in the form of a wave; in essence, it is the vibration of matter. Thus, sound waves cannot travel through a vacuum, but they can travel through matter. Whenever an object vibrates, longitudinal or compression waves are sent forth from the source; this is why vibrations are felt when standing in front of speakers. Similarly, objects struck by sound waves vibrate. 

Converting these causal vibrations into electrical signals is how sound waves are typically detected by human ears, microphones, or other detecting devices. But if the source of the vibrations is enclosed in soundproofing material, then there’s no way to detect the sound. This is precisely why the research stemming from the joint research is so valuable: Vibrations don’t need to be felt; instead, they can be seen. 

“When sound hits an object, it causes the object to vibrate,” says Abe Davis, a graduate student in electrical engineering and computer science at MIT and first author on the new paper. “The motion of this vibration creates a very subtle visual signal that's usually invisible to the naked eye. People didn't realize that this information was there.”

Observing these “subtle visual cues” requires a high-speed camera capable of recording anywhere between 2,000 to 6,000 frames per second. In the experiments involving the potato chip bag, potted plant, and earphones, Davis and his colleague recorded at exactly 5,602 fps. The footage was then interpreted by a unique algorithm they had written and the sound bites were produced.

A later series of experiments revealed that high-speed cameras are not necessary in every circumstance. The meager 60-fps camera found in ordinary smartphones is capable of conveying a relatively decent amount of information; the audio reconstruction cannot compete in clarity with the data pulled from the 5,602-fps sample, but it can identify the gender of the speaker in the room, the number of speakers, and the acoustic properties of the speakers’ voices.

While the law enforcement and surveillance implications of this kind of tech are fairly obvious, Davis and his team are more interested in using the tech to investigating the material and structural properties of objects based on their sound. “We're recovering sounds from objects,” he says. “That gives us a lot of information about the sound that's going on around the object, but it also gives us a lot of information about the object itself, because different objects are going to respond to sound in different ways.” 

Via MIT