Researchers design sounds that can be recorded by microphones but inaudible to humans

June 23, 2017
The sound's frequency is designed by the researchers and transmitted from ultrasonic speakers, completely inaudible to humans but able to be recorded by microphones. Credit: Coordinated Science Laboratory

Microphones, from those in smartphones to hearing aids, are built specifically to hear the human voice—humans can't hear at levels higher than 20 kHz, and microphones max out at around 24 kHz, meaning that microphones only capture the sound we can hear with our ears.

However, researchers at the Coordinated Science Laboratory at the University of Illinois have designed a sound that is completely inaudible to humans (40 kHz or above) yet is audible to any microphone. The sound combines multiple tones that, when interacting with the microphone's mechanics, create what researchers call a "shadow," which is a sound that the microphones can detect.

The team, which includes PhD student Nirupam Roy and CSL Professors Romit Roy Choudhury and Haitham Hassanieh, see many applications for this work. This work won Best Paper Award, titled "BackDoor: Making Microphones Hear Inaudible Sounds," at a leading conference, MobiSys 2017.

"Imagine having a private conversation with someone. You can broadcast this inaudible signal, which translates to a white noise in the microphone, to prevent any spy microphones from recording voices," said Roy, a PhD student in electrical and computer engineering. "Because it's inaudible, it wouldn't interfere at all with the conversation."

According to the researchers, military and government officials could secure private and confidential meetings from electronic eavesdropping or cinemas and concerts could prevent unauthorized recording of movies and live performances.

The signal can also be used to send communication between Internet of Things (IoT) devices, such as an Amazon Echo or Google Home, which would reduce the growing load on Bluetooth, since Bluetooth is primarily how IoT devices communicate. They also foresee that this signal could protect users from unauthorized recording when communicating with voice-activated systems.

"We thought, can we design an application so that when you are actually giving a message, like to an Amazon Echo, no one can record your voice to the Amazon Echo if we're playing this sound?" said Roy. "Voice-activated systems are everywhere, so now it is important to build defenses against attacks that can be launched through your voice."

The sound's frequency is designed by the researchers and transmitted from ultrasonic speakers, completely inaudible to humans but able to be recorded by microphones.

The sound's frequency is designed by the researchers and transmitted from ultrasonic speakers, completely inaudible to humans but able to be recorded by microphones.

The team acknowledges there may be ways to misuse this technology, though they hope that by knowing the problems that can arise, they can build measures to protect against it.

"Inaudible sound jammers, could, for example, affect someone wearing a hearing aid because the internal microphone would pick up that sound," said Roy. "Or, for example, in a bank robbery, someone might be trying to make a phone call to 911, but this sound could jam all the phones trying to make calls."

Like all techniques, inaudible sounds can be used in different ways, but "with this knowledge of how it can be used negatively, we can develop strategies to prevent it," said Roy Choudhury, an associate professor of electrical and computer engineering.

The 's frequency is designed by the researchers and transmitted from ultrasonic speakers, but the microphone—the receiver of the signal—is not altered in anyway. Off-the-shelf microphones will react in the same way to the signal.

"Microphones are in millions of devices, including all of our smartphones," said Hassanieh, an assistant professor of electrical and computer engineering. "And this signal can be received without modifying the , making this technique readily available to interact with the devices around us."

Explore further: Researchers discuss findings on tracking smartphone user habits, activities with ultrasonic beacons

More information: "BackDoor: Making Microphones Hear Inaudible Sounds," synrg.csl.illinois.edu/papers/ … ckdoor_mobisys17.pdf

Related Stories

New research could help build better hearing AIDS

September 29, 2016

Scientists at Binghamton University, State University of New York want to improve sensor technology critical to billions of devices made every year. With a three-year grant from the National Science Foundation, they will ...

Enabling the hearing impaired to locate human speakers

November 26, 2014

New wireless microphones systems developed at EPFL should allow the hearing impaired to aurally identify, even with closed eyes, the location of the person speaking. This new technology will be used in classrooms and conference ...

Recommended for you

Volvo to supply Uber with self-driving cars (Update)

November 20, 2017

Swedish carmaker Volvo Cars said Monday it has signed an agreement to supply "tens of thousands" of self-driving cars to Uber, as the ride-sharing company battles a number of different controversies.

9 comments

Adjust slider to filter visible comments by rank

Display comments: newest first

betterexists
1 / 5 (1) Jun 23, 2017
If so, Make Hearing Aids that modify/amplify for Human Ears!
Dingbone
Jun 23, 2017
This comment has been removed by a moderator.
axemaster
1 / 5 (1) Jun 23, 2017
OK, so they're using intermodulation distortion to exploit the nonlinearities in microphones... so they're playing a pair of INSANELY loud tones at high frequencies. Not very practical in my opinion... And of course, the paper is just a big math proof, which is pointless since everyone knows how to do something like this.

What a bore.
arcmetal
5 / 5 (3) Jun 24, 2017
Can't a microphone be designed to specifically filter out this effect. Then we are back to square one.
nazgand
not rated yet Jun 24, 2017
'"We thought, can we design an application so that when you are actually giving a message, like to an Amazon Echo, no one can record your voice to the Amazon Echo if we're playing this sound?"'
Interesting that the Amazon Echo can somehow hear the voice. I assume this is because it generates the inaudible noise and thus knows what to subtract. If a pseudo-random number generator is used, the eavesdropper could record the noise, guess the state of the RNG generating the noise, and deduce what the human heard.
nazgand
not rated yet Jun 24, 2017
Ultrasound speakers are required to play the sound, and ultrasound microphones exist, which should allow an eavesdropper to calculate the shadow noise to subtract.
gordonhervey1
not rated yet Jun 25, 2017
Without going into a response on every angle, if someone wanted to audibly spy they will know if their microphone source and recording method work. If this technique gets exposed, and we live in the internet era, they could be aware that high frequencies are being used to stop the device from normal operation. A low pass filter negates the effect. A simple phone app or chip could provide a remedy for the next time they try, at worst getting a low pass filter is as difficult as hand wiring a crystal set. Preventing entertainment recordings? For for how long until the technicians' whispers spread.
gordonhervey1
not rated yet Jun 25, 2017
I should have described what I meant in my earlier comment as a loop inline with the microphone element providing a phase reversed signal to restabilise the element, cancelling the high frequency so it's operational.
axemaster
not rated yet Jun 27, 2017
A low pass filter negates the effect

It does not. They are exploiting nonlinearities in the microphone diaphram.

...a loop inline with the microphone element providing a phase reversed signal to restabilise the element, cancelling the high frequency so it's operational.

This will not work because:
- Creating a mechanical loop in the microphone would be very difficult and expensive
- At other frequencies the loop will be in phase and make the effect even worse

The real defense against this attack is that hardly anyone will be gutsy enough to use it, because the high frequencies will have to be loud enough to cause serious hearing damage.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.