University of Michigan researchers develop microphone for smart speakers that doesn’t eavesdrop

‘PrivacyMic’ could be a game changer

A smart speaker.
A smart speaker. (Pexels)

ANN ARBOR – Privacy concerns have long been tied to smart speakers since their microphones can listen to much more than our commands to play a certain song or ask what the temperature outside is.

An estimate 320 million smart speakers are in homes and offices around the world with the capability of hearing everything we say.

But a team at the University of Michigan is hoping to change that.

Researchers have developed a new system that can communicate with a smart speaker without eavesdropping on detectable sound. The system can also listen for the signal that would cause a smart speaker to turn on.

The device, PrivacyMic, uses “ultrasonic sound at frequencies above the range of human hearing,” according to a release. PrivacyMic can hear sounds that dogs and cats can but humans can’t.

The way it works is it pieces together information from ultrasonic sound in our everyday lives to determine when its services are needed.

“There are a lot of situations where we want our home automation system or our smart speaker to understand what’s going on in our home, but we don’t necessarily want it listening to our conversations,” U-M associate professor of electrical engineering and computer science, Alanson Sample, said in a statement. “And what we’ve found is that you can have a system that understands what’s going on and a hard guarantee that it will never record any audible information.”

PrivacyMic is more secure than encryption because it can filter out audible information. Conventional devices can limit who has access to audio data and secure it after it’s recorded, which can leave sensitive information exposed to hackers. With PrivacyMic, that data simply doesn’t exist.

The researchers say the in-home ultrasonic devices could monitor homes of patients in clinical trials or the elderly for signs of problems or how they’re reacting to medications. Ultrasonic devices pick up on 95% of activity in the home, according to the team.

“A conventional microphone placed in somebody’s home for months at a time could give doctors richer information than they’ve ever had before, but patients just aren’t willing to do that with today’s technology,” Sample said in a statement.

“But an ultrasonic device could give doctors and medical schools unprecedented insight into what their patients’ lives are really like in a way that the patients are much more likely to accept.”

The researchers came up with the idea to develop PrivacyMic when they were tasked with classifying audio that was recorded. When analyzing a graph of the data, they realized that much more than audible sound is available.

“We realized that we were sitting on a lot of interesting information that was being ignored,” graduate student in electrical engineering and computer science and first author on a new paper on the research, Yasha Iravantchi, said in a statement. “We could actually get a picture of what was going on in a home or office without using any audio at all.”

Once they had this insight, the team began capturing ultrasonic sound like vacuuming, running dishwashers and other common activities. The ultrasonic signatures were then compressed into smaller files that had key elements of information while they stripped out noise detected by human hearing.

While the device is still in its prototype stage, Sample said that smart speakers would only require minor modifications to have a similar technology implemented in a device. Though potentially years away, the researchers applied for patent protection via the U-M Office of Technology Transfer.

“Smart technology today is an all-or-nothing proposition. You can either have nothing or you can have a device that’s capable of constant audio recording,” Sample said in a statement. “PrivacyMic offers another layer of privacy—you can interact with your device using audio if you choose or you can have another setting where the device can glean information without picking up audio.”


About the Author: