I beg your pardon?! Ear noises (tinnitus), hearing loss or sudden deafness - hearing impairments like these do not only affect older people in Germany. According to estimates by the Deutscher Schwerhörigenbund e.V. (German association for the hearing impaired), more than 5 million people aged between 14 and 19 years were hearing impaired at the end of 2015; 44.6 million of 20 to 60 year-olds and 22.5 million of over 60 year-olds could no longer hear properly. A study by the Institute of Hearing Technology and Audiology at Jade University from 2017 assumes that around 16 percent of all adults in Germany are hard of hearing. The consequences are enormous: every conversation in which one cannot see the other person or background noises occur becomes a torture. Those affected withdraw from social life, suffer more from stress, sleep disorders and dizziness, and their cognitive performance decreases. The risk of developing dementia or depression even doubles.
Intelligent support for the sense of hearing
The good news is that hearing aids can help to remedy this and provide support for the impaired sense of hearing. Provided they are used. But this is precisely the problem: around 80 per cent of hearing impaired people aged between 54 and 75 do not use the little helpers in their ears or do not use them regularly. ”92 percent of people who have a hearing aid but don't use it say that background noise is one of the biggest problems", explains Peter Udo Diehl. And he must know, as he together with his former fellow student Elias Sprengel, whom he met during his Ph.D. between 2013 and 2016 at the Swiss Federal Institute of Technology in Zürich - ETH Zürich, made this problem their profession - or rather the solution to it. “We were both fascinated by MedTech and AI“, explains the CEO, “Elias was already working on sound classifications at the ETH and on machine learning at Google. During my master's degree in cooperation with University College London (UCL), I conducted studies on audio perception and developed AI strategies for companies at McKinsey & Company.“ The professional interest in hearing technology and hearing matched, the personal interest in any case - the founding of a joint start-up seemed to be the logical consequence. So they both left their jobs and brought the Audatic start-up to life in January 2018. Since then, the international team of 10 has been writing a success story. The reason is the groundbreaking product: software in which state-of-the-art algorithms use neural networks to recognize sounds and separate noise from speech. The result is a personalized sound environment in which even the hard of hearing can hear again. For: “Hearing aids work excellently in quiet environments and 1:1 settings, but if you add street noise or are in a crowded restaurant, these background noises make understanding difficult“, according to Diehl, who has studied at the TU Dortmund, the Humboldt University of Berlin and TU Berlin and University College London (UCL) among others. “We use current methods of deep learning to filter out the background noises and restore the 1:1 settings.“
Small Artificial Intelligence on a large scale
This is not the only innovation. While the neural networks of AI systems are usually used and trained on large servers, the Berlin team focuses on the motto "Small, but powerful". ”Last year we worked on moving our software from the server to the laptop and smartphone“, says Diehl, who together with his co-founder Sprengel recently landed on the Forbes 30 under 30 list. “Currently, many neural networks are calculated with computer accuracy of seven digits after the decimal point. But they don't have to be so precise. The nice thing is: even if the accuracy is reduced by 80 - 90, the functionality remains the same. AI systems are thus just as robust against errors as our brain.” The team has already succeeded in bringing the Audatic software to the smartphone. It lies on the table, cleans the environment of background noise and sends the intelligible signals to the hearing aid in real time. ”However, it will take another two to three years before the technology sits in the hearing aid”, Diehl estimates. He also envisages a further application of Audatic technology in the medium term. ”The worlds of hearing aids and hearables like the wireless headphones AirPods are merging more and more”, the young entrepreneur knows, ”today you can put your iPhone on the table and stream music to AirPods. Maybe it will soon be possible to reduce background noise in noisy restaurants by 50 percent. And if Klara speaks more quietly at the end of the table, I can make her voice louder. Or I can turn off the traffic noise in the city, but amplify the chirping of birds and make my day more pleasant.“
Berlin: Centre of HealthTech solutions
How such a hearing experience is not only good for our ears, but also positively influences our quality of life, is confirmed by users after the first tests of the Audatic software: ”We have tested 50,000 audio examples with people without hearing loss and also with a handful of hearing aid wearers“, says Peter Udo Diehl and adds: ”We are currently in discussions with the ENT department of the Berlin Charité Hospital to conduct a clinical study.“ If everything goes according to plan, the team hopes to be able to start in two to three months. At any rate, the whole thing should not fail because of the cooperation partner: “Charité was enthusiastic about the partnership“, reports the AI expert, “they have the drive and are looking forward to further developing our product with us.“ For the start-up founder, the popularity of the renowned university hospital is not the only sign that Berlin has been chosen as the right location. Also, Audatic's proximity to research institutions such as the Technical University of Berlin and the Bernstein Center for Computational Neuroscience Berlin has already paid off. Not only Diehl himself did his master’s degree there (and discovered his penchant for Artificial Intelligence in connection with human hearing); also one of the first trainees of the start-up came from the program. ”Deep learning for mobile devices has to be super efficient and simultaneous, so we need proximity to the universities," Diehl is convinced, "we need proximity to the best talents and the best technology. Both go hand in hand. Knowledge comes with the right people.“ It was above all these "right people" who came out in favour of Berlin from the very beginning - literally: “When we founded the company we had three main candidates: London, Berlin, Zürich“, says the native of Gera, “we had a wish list of potential employees from all over the world and asked them where they would like to move to. An AI start-up ultimately lives or dies with its talents. The choice was clearly Berlin.”
It seems to have been the right decision: in November 2018, Audatic was the best HealthTech start-up from among 30 submissions and received the "Digital Health Award" from the Berlin Senate Department for Economics, Energy and Public Enterprises. A prize which not only made the young company even better known in one fell swoop, but can also help to find investors and cooperation partners. While the financial aspect is currently well covered thanks to private investors, research funds and the European Regional Development Fund, the latter in particular are currently in demand. ”We would like to cooperate with one or two hearing aid manufacturers," says Peter Udo Diehl, explaining the next steps, "but we are also talking to telephone manufacturers. After all, it can be challenging for people with normal hearing to suppress background noise at the airport or on the train.“
Whether hearing-impaired persons or persons with normal hearing, one thing seems certain: Audatic will have a lot more to say in the future...