thanks to AI, people with auditory neuropathy will soon have adapted prostheses

Until now, there has been no effective hearing aid for people who have difficulty hearing in noisy environments. A team from the Pasteur Institute launched a study on volunteers with the help of artificial intelligence.

Imagine the sound of a poorly tuned radio. That’s pretty much the vibe a person hears Suffering from auditory neuropathy in an airplane taking off. “It’s a very particular profile, explain Grégory Gérenton, engineer at the Center for Research and Innovation in Human Audiology (Ceriah). They’re going to mix up syllables, they’re going to mix up a whole bunch of sounds.”

Ceriah, hosted by the Pasteur Institute, has just launched a study to develop a hearing device equipped with AI and adapted to this pathology, which represents 10% of deafness. “In everyday conversation, they manage to catch up because they have the context, because they have lip reading. It gives them clues.” continues Grégory Gérenton. “Many situations are critical for them. Even understanding on the street, with cars passing by, is going to be very problematic, just like in restaurants or during meetings.”

Existing hearing aids amplify and filter sounds but this does not help neuropathic patients. To meet their needs, researchers will first precisely define what the blockages are, thanks to tests carried out on volunteers. For example, we make them listen to sentences distorted by noise or chopped up.

Prosthetics equipped with AI

Researchers from the University of Lorraine will then test algorithms on around forty volunteers, mostly young people who are still working. Half of them hear well, the other half suffer from auditory neuropathy. “We will see them every year and each year we will offer them speech enhancement algorithms, which we will have developed with our partners and which, we hope, will help them to better understand speech,” detailed Céline Quinsac, project manager at the Institut Pasteur research center.

“For example, if we make them listen to a text in year one, they will perhaps understand 50% of the words. We hope that in year 2, they will understand 70%, etc….”

Céline Quinsac, project manager at the Institut Pasteur research center

at franceinfo

The CEA, the Atomic Energy Commission, must then develop hearing aids equipped with AI capable of separating speech from ambient noise, using algorithms tested on volunteers.