Cochlear implants (CIs) are among the most successful neural prostheses : they are able to replace a full sensory organ and CI users can follow conversations again.
Our measurements show, that CI users understand speech in steady state background noise surprisingly well, but their performance drops in modulated interference, a condition where normal hearing listeners benefit from a masking release.
To better understand neuronal coding in the implanted ear, we develop models of the electrical excitation of the auditory nerve and we analyse the quality of coding with various methods. With an automatic speech recogniton system we quantify how well speech is coded in noise and we use a binaural model to analyse how precise interaural time differences are coded by contemporary coding strategies.
Our modeling efforts are supposed to pre-screen novel coding strategies before they are tested in lenghtly patient studies, which should speed-up new developments considerably.
