How do variations in cochlear implant map settings alter the spectral representation and the signal dynamics? – A simulation study

Authors: Mona Kirstin Fehling1, Johannes Burkart1, Benedikt Kramer1, Nicole Rotter1, Angela Schell1

1Department of Otorhinolaryngology, Head and Neck Surgery, University Hospital Mannheim, Medical Faculty Mannheim of Heidelberg University

Background: After cochlear implant (CI) surgery, repeated fitting of the CI speech processor is required to ensure good hearing and speech intelligibility. This fitting is time-consuming and exhausting for patients but crucial as the resulting speech processor settings (map) are highly individual. This study systematically investigates the impact of different map settings on spectral representation and signal dynamics to provide insights that could enhance this process.

Methods: A MATLAB vocoder was used to simulate the spectral representation of different CI map settings (variations in IDR, T- and C-Levels). To investigate the fine-grained spectral representation crucial for speech intelligibility, a recording of 9 subsequent phonemes (ADA-AFA-AHA-AKA-AMA-ASA6-ASA9-ASCHA3-ASCHA5), which was chosen for its structured phonetic content, served as input and reference. The so-generated vocoder signal was then analyzed regarding bark band-specific energy distribution, spectral representation, and signal dynamics.

Results: Quantitative analysis showed variations in bark band-specific energy distribution, with some settings increasing energy concentration in bands crucial for speech intelligibility. Analysis of signal dynamics indicated a reduction in dynamic range under narrower IDR settings, affecting the signal’s amplitude fluctuations. Spectrograms confirmed these differences in spectral peaks and temporal dynamics across settings.

Conclusions: These findings demonstrate that small changes in Map settings can significantly affect spectral resolution and signal dynamics, critical for frequency discrimination and sound intensity perception. A deeper understanding of how variations in CI maps affect these aspects will facilitate CI fitting by allowing for automatic generation of map recommendations based on input-output analysis, as already established in hearing aid acoustics. This will prospectively make fitting more efficient for patients and ensure better outcomes.