Comparative evaluation of cochlear implant coding strate- gies via a model of the human auditory speech processing

Traditional cochlear implant (CI) coding strategies present some informa- tion about the waveform or spectral features of the speech signal to the electrodes. However, neither of these approaches takes the cochlear travel- ing wave or the auditory nerve cell response into account, though these are given in acoustic hearing. Therefore, a new CI coding strategy based on an auditory model including the above mentioned properties of the healthy cochlea was evaluated and compared with an n-of-m-coding strategy, in which n electrodes out of m possible electrodes are stimulated in each stimulation cycle. The selection of the n electrodes is based on the n highest spectral maxima of the momentary signal. Simulated electrical output of both CI coding strategies served as input to a model of the electrically stimulated auditory system, which consisted of an auditory nerve cell population. The nerve cells generated delta pulses as action potentials in dependence on the spatial and temporal properties of the electric field produced by the electric stimuli. This model is used to predict CI user performance in terms of speech intelligibility and pitch discrimination for both coding strategies. Furthermore, an additional model of normal hearing is presented, the output of which is compared to the neural representation resulting from the modeled CI stimulation. We will show under which circumstances and to what extent an auditory model based coding strategy may outperform a traditional CI speech coding algorithm.