Siri, the iPhone’s voice-activated assistant, may soon be much easier to communicate with in noisy environments, thanks to new sensory technology from Duke University that solves the problem of crowd noise.
In a crowded room where voices are coming from every direction, the human auditory system is naturally good at focusing on a single voice while filtering out background chatter. Computers are not good at this, which might explain why it can be so frustrating for iPhone users to make themselves understood to Siri.
According to an article that appeared in the August 10, 2015 edition of the Proceedings of the National Academy of Sciences (PNAS), engineers at Duke University’s Pratt School of Engineering have developed a sensor that uses meta-materials — the combination of natural materials in repeating patterns to achieve unnatural properties — and compressive sensing to help computers determine the direction of a sound and extract it from the surrounding background noise. Once miniaturized, the sensor could have applications in many voice-command electronics and medical sensing devices that use waves, like ultrasound, hearing aids, and cochlear implants.
“We’ve invented a sensing system that can efficiently, reliably, and inexpensively solve an interesting problem that modern technology has to deal with on a daily basis,” said author Abel Xie, a PhD student in electrical and computer engineering at Duke. “We think this could improve the performance of voice-activated devices like smart phones and game consoles while also reducing the complexity of the system.”
According to an announcement from Duke, the proof-of-concept device looks like a thick, plastic, pie-shaped honeycomb split into dozens of slices. The honeycomb’s openings all look the same, but their depth varies from hole to hole. This gives each “slice” of the honeycomb pie a unique pattern.
“The cavities behave like soda bottles when you blow across their tops,” said Steve Cummer, PhD, professor of electrical and computer engineering at Duke. “Like the amount of soda left in the bottle, the depth of the cavities affects the pitch of the sound they make, and this changes the incoming sound in a subtle but detectable way.”
When a sound wave reaches the device, it becomes slightly distorted by these cavities. That distortion has a specific signature, depending which slice of the pie it passed over. After being picked up by a microphone on the other side, the sound is transmitted to a computer that is able to separate the jumble of noises based on these unique distortions. The researchers tested their invention in multiple trials by simultaneously sending three identical sounds to the sensor from three different directions. The sensor was able to distinguish between sounds with a 96.7 percent accuracy rate.
While the prototype is six inches wide, the Duke team believes it could be scaled down and incorporated into the devices we use on a regular basis. And because the sensor is made of plastic and does not have any electric or moving parts, it is efficient and reliable. The researchers think the concept may also have applications outside the world of consumer electronics–it could possibly be combined with any medical imaging device that uses waves, such as ultrasound, to improve current sensing methods, and also create new ones. The Duke team says that with additional research, it should also be possible to improve the sensor’s sound fidelity and increase its functionality for use in hearing aids and cochlear implants.
Source: Duke University
Image credit: Steve Cummer, Duke University; © Norman Chan | Dreamstime.com