Pulse Audition: enabling the hard of hearing to hold conversations in noisy environments

People who are hard of hearing are not deaf, but instead suffer from hearing loss, which can manifest itself in a number of ways: difficulties speaking on the phone, or hearing things like the doorbell, birdsong or the sound of cars as they approach. It can also make it impossible to follow conversations in noisy environments, such as busy streets, big shops or parties. This is something that affects 430 million people worldwide, fewer than 20% of whom wear auditory implants.

The project Pulse Audition, which was awarded the Grand Prize at the i-PhD innovation competition in 2021, is not aiming to compete with the five players dominating this market. “We are focusing exclusively on noisy environments, which can be a real issue for the hard of hearing”, explains Manuel Pariente, one of the key figures behind the project Pulse Audition. “They might be with friends or family, but communicating with them is very difficult. This makes them frustrated and sad, and there is a tendency for them to withdraw or even to decide not to go at all.” 

“Our priority is improving quality of life for the hard of hearing”

This was something which Manuel Pariente and his future associate Thibaud Moufle-Milot experienced at family get-togethers. “I would be sitting next to my grandpa, and I would turn around to speak to him, but he wouldn't be able to understand me”, recalls Thibaud. “Either that or I would need to ask everyone else to be quiet.” 

This was partly the inspiration behind Pulse Audition, which should result in the launch of a medtech startup before the end of 2022. If the hard of hearing are able to hold conversations in noisy environments, then they will be able to reengage with the social interactions that make life enjoyable. 

What the two creators have come up with is a pair of glasses which, although they might look ordinary, are fitted with microphones capturing sound through 360°, sensors which detect which way the person is facing, and technology for transmitting sound through bone conduction via the temples of the glasses, keeping the ear canal clear.

Invisible, but vital, artificial intelligence and signal processing software records all sounds in order to map the “auditory scene”: who is speaking? Where from? How far away are they? When? This map is then cross-referenced with facial orientation in order to determine the conversation that the wearer of the glasses wants to follow. This conversation is then “cleaned up” of any noise, echoes or reverberations, delivering optimal clarity.

The creators: one IT researcher and one MedTech engineer

In order to develop this “speech enhancement” - the term used by experts in the field - three years had to be spent developing and integrating artificial intelligence, signal processing and Bayesian statistics (an approach centred around deductions based on real-life observations) tools. Because the glasses are fitted with several microphones, signal processing is multi-channel. The device must attain a latency time (the delay between the input sound and the amplified sound) shorter than 10 milliseconds in order to rival commercial implants and to ensure a smooth experience for users. “Above 15 milliseconds, this delay becomes noticeable, which would be a real nuisance”, explains Manuel Pariente.

Despite the progress that has been made, it will still take several years for the two innovators behind the project to develop their product, drawing on their complementary skillsets. Manuel Pariente is an IT researcher; Thibaud Moufle-Milot, a friend of his from high school, is a medtech engineer and business developer. He will be bringing to the table his experience in entrepreneurship: “You need to understand the market, which is divided up into different auditory issues and patient profiles: some wear devices, while others don’t”, he explains. “Actual glasses have to be designed, which requires skillsets external to IT. This involves equipping the glasses with software, sensors and batteries, and doing it discreetly enough that they look like regular glasses.”

Even more importantly, Pulse Audition must be prepared to take on board feedback from potential users. Will they finally be able to understand conversations they’re interested in? What will they think of the sound quality? In noisy environments, will they notice a clear difference compared to the best implants on the market?

The startup is set to be launched officially in October 2022, a short while after leaving Inria Startup Studio, where it has been hosted, supported and funded for a year. “These twelve months have given us the confidence needed to get started and the security to do it”, says a happy Manuel Pariente. “Instead of having to speed things up in order to create, we’ve had the time to progress, to make mistakes and to learn lessons from them”, adds Thibaud Moufle-Milot. They don't yet have a clear idea of when the spectacle hearing system will be available on the market: “Not in 2023, that would be premature. But within three years, that much we know.” Watch this space...

• Manuel Pariente - Winner of the i-PhD 2021 competition for the project Pulse, Génération Deeptech Bpifrance, 29/3/2022.
• The psychological implications of hearing impairment, EDP Audio, 14/11/2015.
• Technology for the elderly: a new challenge for research and development in France? Inria, 2/11/2021.