The CherISH network

Project 1: Tactile assistive system for sound localization in cochlear implant users

Host Institution: Eberhard Karls Universität Tübingen, Tübingen, Germany.
The project will be conducted in close collaboration between the Center for Magnetoenzephalography (MEG) and the Machine Learning in Sciences (part of the Excellence Cluster Machine Learning Tübingen and the Tübingen AI Center) in Tübingen, Germany.

In the Project 1, an assistive system (AS) for sound localization in cochlear implant (CI) users is developed. The AS decodes sound directions using artificial neural networks. Based on decoding results, tactile cues will inform CI users about sound directions. The AS is tested in a lab environment and in controlled open-field experiments. It will allow for continuous training of spatial hearing in everyday situations. Neuroplastic changes during the restoration of spatial hearing are studied with EEG. Physiological activity will be compared to the performance of deep neural networks (DNNs) in order to identify problems of spatial hearing and design appropriate training procedures. 

Our partners for this project:

• Nemo Lab/Spatial Sound Institute, former 4D Sound, Hungary [Secondment 1]
• UCBL, Lyon Claude Bernard University, France [Secondment 2]
• RCNS, TTK Research Centre for Natural Sciences, Hungary
• BME, Budapest University of Technology and Economics, Hungary