The LTS2-Acoustics has developed a binaural localization algorithm for hearing aid purpose. Strong technical constraints have restricted the possible signal processing methods available for such a task, limiting the localization performance. New hearing aid systems will incorporate more processing power and memory storage, which should allow more powerful algorithms to be developed.

Using a huge database of measurements acquired in different acoustic environments, the student will be asked to study and implement new methods for localization targets. From existing techniques related to Fourier analysis and correlation, the student will propose a novel approach to clean and process the acoustic signals, while favoring the simplicity of the computations in a view of future implementations on embedded devices.

Profile: Electrical engineering, computer science

Prerequisites: Electroacoustique (BA5) or Audio Engineering (MA1), basic knowledge with Matlab

Content: Acoustics, Signal processing, Matlab implementation

Reference: http://actu.epfl.ch/news/enabling-hearing-impaired-to-locate-the-speaker/