AI Can Hear You Now

An onboard neural network allows this hearing aid to work naturally, like a brain, in amplifying relevant sounds for greater understanding.

Nick Bild
a month agoMachine Learning & AI
(📷: Oticon)

The National Institutes of Health note that nearly twenty-nine million people in the United States alone could benefit from hearing aids. Considering the prevalence of hearing issues, and the dramatic impact those issues can have on one’s quality of life, it should come as no surprise that advancements are being made in devices that better treat hearing loss.

One of the more notable advancements is being made in the Oticon More line of hearing aids developed by Oticon. Oticon More hearing aids process audio onboard with deep neural networks that have been designed to give the brain more of the relevant information it needs to make better sense of sound, rather than taking a more conventional approach of simply amplifying all sound. Oticon claims that their device reduces listening effort so that the wearer can remember more of what is being said.

At the heart of the Oticon More is the Polaris platform, which consists of a digital signal processing chip for audiology processing, a front-end chip that samples the microphone signals, a near-field-magnetic-induction (NFMI) radio, and a Bluetooth Low Energy radio.

Audio signals are captured either by the onboard microphones, or through a Bluetooth connection to the instrument. The audio signals are processed through a filtering pipeline that prepares them for the next step, the deep neural network. The neural network processing causes meaningful sounds to stand out from unimportant background noises similar to the way our brains naturally focus on the most relevant sounds. Next, an optional filter can be enabled to reduce wind noise. Finally, the signal is amplified — within safe limits — and then output. The NFMI radio is used to keep audio synchronized between the pair of devices.

The neural network was trained on twelve million real-world sound samples. Oticon believes that in learning by example to recognize important noises, the algorithm will be able to decipher the intricate details of relevant sounds in a way similar to the brain, which also learns by example.

Wireless connectivity allows the Oticon More to stream audio from devices such as smartphones, tablets, and televisions, in addition to collecting environmental sounds. Built-in rechargeable batteries can power the hearing aids for a full day with a three hour charge time. A companion phone app is available that allows users the ability to configure the devices as needed.

If Oticon More lives up to the promises, I think we will be hearing a lot more (pun intended) about Oticon in the future.

Nick Bild
R&D, creativity, and building the next big thing you never knew you wanted are my specialties.
Related articles
Sponsored articles
Related articles