Researchers in the US have developed a system that can inform a smart home — or listen for the signal that would turn on a smart speaker — without eavesdropping on audible sound.
Microphones are among the most common electronic sensors in the world, with approximately 320 million people listening to our commands in the world’s smart speakers.
However, such technology can also hear everything else. To challenge this, the ‘PrivacyMic’ device, developed by a team at the University of Michigan (U-M), uses ultrasound at frequencies above the range of human hearing.
Running dishwashers, computer monitors and even finger clicks all generate ultrasonic sounds at a frequency of 20 kHz or higher.
The system aggregates the ultrasonic information in its environment to determine when a user needs its services and to sense what is happening around them.
The researchers have shown that it can identify household and office activities with greater than 95 percent accuracy.
“There are many situations where we want our home automation system or smart speaker to understand what’s going on in our house, but we don’t want it to listen to our conversations,” said Alanson Sample, UM associate professor of electrical engineering and computer science.
“And what we found is that you can have a system that understands what’s going on and a hard guarantee that it will never record audible information.
” The researchers said PrivacyMic can filter audible information directly on the device, making it more secure than encryption or other security measures that take steps to protect audio data after it’s been recorded or restrict who can access it.
Those measures can all make sensitive information vulnerable to hackers, but with PrivacyMic, the information simply doesn’t exist, the team said.
The idea behind PrivacyMic started when the team classified previously recorded audio. Looking at a visual graph of the data, the researcher found that audible sound was only a small fraction of what was available.
“We realized that we were sitting on a lot of interesting information that was being ignored. We could really get a picture of what was happening at home or in the office without using any audio,” said Yasha Iravantchi, a graduate student of electrical engineering and computer science.
With this knowledge, Using a laptop and an ultrasound microphone, the team then captured audio from brushing teeth, flushing toilets, vacuuming, running dishwashers, using computer screens, and hundreds of other common activities, then compressing the ultrasound signatures into smaller files that contain important bits of information, while reducing noise.
within reach of human hearing — similar to an ultrasonic MP3 — and built a Raspberry Pi-based device to listen to. According to the team, the device, which can be set to filter out speech or remove all audible content remove, accurate 95 percent of the time very common activities.
The team also conducted a trial where study participants listened to the audio collected by the device and found that no participant could understand human speech.
While the device is a proof of concept at this stage, Sample says implementing similar technology in a device like a smart speaker requires only minor tweaks: software that listens for ultrasound and a microphone that can pick it up, which is cheap. and available.
“Smart technology is an all-or-nothing proposition these days. You can have nothing or you can have a device that can constantly record audio,” Sample said.“PrivacyMic offers another layer of privacy:
you can interact with your device using audio if you want, or you can have another setting where the device can provide information can collect without picking up audio.”