Robert Rehr

My research focuses on machine learning-based algorithms that assist people with hearing loss in their daily lives, such as removing noise to help them focus on sounds that are important to them. This includes noise reduction algorithms, but can also include algorithms that identify which sounds are most important to the user.

My interest in the field of hearing care stems from my fascination with music and music production. These hobbies led me to the “Hearing Technology and Audiology” program, which was recommended by the parents of a friend of mine. This program, which is offered at the Jade Hochschule and the University of Oldenburg, all aspects of audio with a special focus on hearing care. The opportunity to use my skills to solve a problem that affects many people is what drew me to the field of hearing care.

After completing my Ph.D., I joined Oticon as a developer working on noise reduction algorithms. During this time, I had the opportunity to work on a challenging project and deepen my understanding of machine learning, which I genuinely enjoyed. Fortunately, the recently established AI focus area at Eriksholm Research Centre allows me to continue this kind of work.

I’m motivated by solving challenging problems that matter to people, and I love learning and exploring new things.

I hope to contribute to a better understanding of how neural network-based algorithms work, e.g., for speech enhancement. My goal is to provide a deeper understanding of such machine learning-based algorithms, to show their capabilities as well as their limitations, so that we can harness the power of machine learning more efficiently.

I am a member of a stand-up paddling (SUP) club in Copenhagen where I enjoy paddling in the canals of Copenhagen and Amager Strand with other stand-up paddlers. Since 2023, I have participated in the Danish SUP Tour, which is a series of competitions organized in Denmark. I also enjoy playing the piano and spending time with my friends.

The rise of deep learning has been one of the most disruptive and exciting innovations of recent times. This technology offers new solutions to problems that we have been trying to solve for many years, often with better performance than previous approaches. One aspect that stands out to me is the role of automatic differentiation frameworks, which have made deep learning more accessible and are now being used across different scientific disciplines.

The number of scientific papers is increasing every day. While it is great to see so much new knowledge being generated, the sheer volume of papers has made it nearly impossible for humans to process. To make this knowledge accessible, I hope we develop tools to efficiently search and summarize this growing body of research.