KISH – AI-based selective hearing via headphones

Hear What's Important

"A cyclist who's in traffic with headphones. The car approaching him from the side. He doesn't see it - and he can't hear it either. When it collides, he has no chance." He could still be alive - with the help of an intelligent technology that makes an approaching car more audible to the user. Behind this vision is an intelligent headphone system that helps its users recognize concrete sound events, separate them from each other, and selectively amplify or suppress them.

The human ear is always on and sensitive in all directions. While you can close your eyes to avoid visual stimuli or focus on objects with your eyes, the possibilities with your ear are limited. It is often too loud in the environment, certain noises disturb or we consciously want to concentrate on specific acoustic information in the environment.

The project "KISH - KI-based selective hearing for headphones" aims to develop intelligent hearables that support the user in making certain sound events more audible in a complex acoustic environment, while allowing the targeted suppression of sound events perceived as disturbing. Innovative AI sound separation algorithms ensure that the sound sources in the user's acoustic environment can be recognized analyzed and processed separately. For the listener, this creates a sound experience tailored to individual needs by allowing defined acoustic objects to be amplified, attenuated or modified.

Based on the extensive previous experience gained in industrial and public projects, including detection and separation of sound sources as well as audio rendering and active noise control, Fraunhofer IDMT takes the leading role in the project and is responsible for developing recommendations for the further development of the ecosystem "selective hearing with AI support". Together with the project partners, we will validate the technical feasibility and economic relevance of this concept. The declared goal is to gain detailed knowledge about future, concrete marketing opportunities, to define the appropriate target market, to develop market entry scenarios and to acquire strategic partnerships.

Detect hazard noises - The acoustically transparent headphones detect hazard noises in the environment and thus warn the user.

Enable communication - The acoustically transparent headphones detect speech and enable communication.

Detect warning signals - The acoustically transparent headphones can react to warning signals and alert the user to dangerous situations.

Partners

Subcontracted Partners

Associated Partners

Advisory Board

Sponsors

Duration

April 2019 – October 2019 (competition phase)