A lightweight earpiece technology promises to meet or beat the performance of the best premium noise-canceling headphones without blocking the ear canal or covering people’s ears like heavy earmuffs. The new noise-cancellation device leverages the not-so-secret fact that wireless network signals can travel 1 million times faster than sound waves.
Commercial noise-canceling headphones and earphones have internal microphones to detect incoming sounds so that their digital signal processors (DSPs) can compute an appropriate antinoise signal for canceling each incoming sound. But having microphones located so close to the ears means commercial headphones get just tens of microseconds to process incoming sound, calculate an antinoise signal, and play it through their speakers before the sound reaches the ears. Such a tight computing deadline limits the performance of most active noise-cancellation technology, setting an upper limit on the frequencies it can actively cancel. This requires many headphones to shield the wearer’s ears with sound-absorbing materials for higher frequencies.
“The main hardware challenge is to get rid of any sound-absorbing material, and instead design a wireless forwarding system that runs very fast to provide the maximum ‘head start’ to the ear piece,” says Sheng Shen, a Ph.D. student in electrical and computer engineering at the University of Illinois at Urbana-Champaign.
Shen and his colleagues plucked the usually embedded microphones out of the earpiece to make a noise-cancellation system—called MUTE—with physically separated components connected through a wireless network. External microphones are placed closer to noise sources; you might put one by, say, your office door if people tend to chat outside. Because the wireless signal arrives well in advance of the sound waves coming through the door, this gives the MUTE system more time to process incoming sounds and calculate the corresponding antinoise signals. The time advantage of having early information about “future sounds” can mean having several milliseconds—hundreds of times longer than the tens of microseconds available to conventional headphones—to perform the necessary noise-cancellation computations.
Having the luxury of more time for the DSPs to do their work means the system can counter more complex sounds and noises, Shen says. Early lab testing suggests that the new technology could exceed the performance of a premium noise-canceling headphone such as the Bose QuietComfort 35 Smart Headphone, as described in a paper presented at the ACM SIGCOMM Conference held in Budapest, Hungary, from 20–25 August 2018.
Imagining a future MUTE device.Illustration: University of Illinois at Urbana-Champaign
The first version of the MUTE system is a crude lab prototype consisting of cheap microphones, speakers, and DSP boards. But given the initial promise shown by the technology, the researchers envision developing MUTE into a wearable device in the form of a behind-the-ear frame with a hollow, cylindrical speaker snuggled into the ear. Unlike ear-covering headphones or even ear-blocking earbuds, the MUTE earpiece could keep the ear canal mostly unplugged.
“The advantage is that such hollow devices are much more comfortable to wear, and even healthier, due to continuous air circulation through the ear canal,” says Romit Roy Choudhury, a professor of electrical and computer engineering at the University of Illinois at Urbana-Champaign.
Several rounds of testing pitted the early-version MUTE system against the premium Bose headphones that use both active noise cancellation and passive soundproofing material. From the start, MUTE proved it could perform active noise cancellation at sound frequencies between 0 and 4 kilohertz, compared to the Bose headphones’ active noise-canceling range of 0 to 1 kHz. (The Bose headphones rely on passive soundproofing material to dampen sounds in the 1- to 4-kHz range.)
An upper limit of 4 kHz includes much of the frequency spectrum that contains human voices and music. The researchers anticipate MUTE will be able to handle more of the spectrum containing speech and music along with other sounds associated with frequencies above 4 kHz—such as glass-breaking or door-slamming sounds—if they use faster digital signal processors.
Within the lower 1-kHz range, MUTE also outperformed the Bose noise-cancellation technology by 6.7 decibels on average in terms of dampening sound intensity (loudness).
When MUTE alone faced the combination of the Bose headphones’ active noise cancellation and its passive soundproofing material, the Bose headphones’ overall noise cancellation proved slightly better by almost 1 dB. But when the passive soundproofing materials were added to the MUTE system, it outperformed Bose once more, by 8.9 dB.
The additional processing time enjoyed by MUTE also allowed the researchers to design software algorithms that can more seamlessly adjust the noise-canceling signals in situations involving two or more very different noises. For example, the MUTE software can rapidly switch between different sound profiles for the intermittent sounds of a human voice versus the constant background hum of machinery.
The remote wireless microphone (black and red device connected by three wires) and the wireless relay. The relay sends a 900-megahertz analog FM signal to the noise-canceling device.Photo: University of Illinois at Urbana-Champaign
But the external microphone system also comes with some practical limitations for future users. One challenge is that MUTE currently works only in natural indoor environments with a single dominant noise source, such as a person chatting on the phone or music playing from an audio speaker. The researchers want to expand MUTE’s noise cancellation to multiple noise sources with the help of several different wireless microphones—each acting as an Internet of Things relay—and source separation algorithms that can distinguish between noises.
Another challenge is that the external microphones must always be closer to the noise it’s canceling than the person wearing the MUTE earpiece. That may require several microphone relays scattered around a person to intercept incoming sounds from all directions.
The reliance on external microphones relays also makes MUTE much less portable than traditional noise-canceling headphones. The system may work best in more stationary scenarios that involve working at the office, sleeping at home, or working out in a gym, Shen says. It’s unlikely to prove practical for joggers who want to enjoy the noise-canceling benefits unless MUTE relays are scattered all around a large area.
Still, Shen is excited about the possibility of putting the noise-canceling technology into a personal tabletop device that might act as a smart noise-cancellation assistant similar to the Amazon Echo or Google Home systems. That might prove the most portable version in combination with the user wearing the hollow earpiece.
Privacy and security concerns often go hand in hand with such smart devices. The researchers addressed privacy concerns in their paper by noting that the microphone relays in their prototype system use analog signal processing and are not designed to either record incoming sounds or hold acoustic samples. But they also acknowledged the possibility of a “malicious eavesdropper” intercepting some of the wireless relay signals and potentially being able to overhear things from far away—a potential threat that could be mitigated with the right techniques involving power control and beamforming.
Another possible implementation could have multiple microphone relays to perform noise-cancellation duty for many different people in a certain area, using digital signal processing offloaded to cloud computing servers. That “public edge service” version of MUTE could potentially serve call centers or other relatively noisy environments where people remain mostly in one spot.
A third “smart noise” version could also perform a public service for neighboring people. This could involve a microphone relay placed directly on top of noisy construction machinery or lawnmowers so that the relay could broadcast the sound wirelessly to anyone with a noise-canceling earpiece.
The relative luxury of having “future sound samples” from the microphone relays could provide enough time and data to someday enable smarter AI-driven sound applications for earphones based on machine learning-algorithms, says Shen. That could open the door for even more intriguing applications such as sound-enhanced augmented reality games or apps.
But for now, Shen and his colleagues plan to focus on making their noise-cancellation device into something resembling more of a commercial product and less of a lab experiment: “We want to go from our current prototype to a device that looks like a somewhat wearable product behind human ears.”
Jeremy Hsu has been working as a science and technology journalist in New York City since 2008. He has written on subjects as diverse as supercomputing and wearable electronics for IEEE Spectrum. When he’s not trying to wrap his head around the latest quantum computing news for Spectrum, he also contributes to a variety of publications such as Scientific American, Discover, Popular Science, and others. He is a graduate of New York University’s Science, Health & Environmental Reporting Program.