Contribution
NeuroSensEar: Neuromorphic acoustic sensing for the high-performance hearing aids of tomorrow
* Presenting author
Abstract:
Hearing aids and cochlear implants are familiar to most people. Although they can significantly improve the lives of people with severe hearing impairment or deafness, only 41% of those affected in the EU use such support. The reason for this is that, despite considerable progress in terms of performance and user-friendliness, there are still difficulties in understanding speech and adapting the devices. NeuroSensEar addresses precisely this weakness of conventional hearing aids.Our goal is to address the limitations of current hearing aids: improving speech understanding in noisy environments and enabling devices to adapt to new situations through continuous learning. To overcome these challenges, NeuroSensEar draws inspiration from biological information processing, specifically the hearing mechanisms of insects. Our system is based on (i)the MEMS cochlea, a neuromorphic acoustic sensor with integrated signal processing, (ii)neuromorphic and application-specific circuits for low-power and small system implementation(iii)artificial neural networks for acoustic scene analysis and (iv)control for automatic adaptation to the acoustic environment and the user. The performance of the system is tested using virtual acoustic environments. First results of the project will be presented, including speech processing under noisy conditions and small ANNs for ASC.