Speech understanding in quiet and in noise with the CIS speech coding strategy (MED-EL Combi-40) compared to the multipeak and spectral peak strategies (Nucleus)

This study compares sentence understanding in quiet and in noise with 3 different speech coding strategies for cochlear implants. The results show that the spectral-peak (SPEAK) and continuous-interleaved-sampling (CIS) coding strategies, based on spectral signal analysis, allow for better speech understanding in quiet as well as in noise, than the multipeak (MPEAK) coding strategy, which relys on speech feature extraction. In the intrasubject comparison of the MPEAK and SPEAK strategies, the SPEAK coding strategy provided a considerable improvement in quiet and in noise for the majority of patients using the Nucleus 22 Mini-implant. In the intersubject comparisons, the mean results in noise with the CIS strategy were superior to both the MPEAK and the SPEAK strategies. The difference was greatest for the most difficult tests in noise. Understanding in noise was least reduced for the CIS strategy. Understanding in quiet was not significantly different between the CIS and the SPEAK strategies; both strategies were significantly better than the MPEAK strategy in quiet. These results are still preliminary, due to the relatively small number of patients and the great inherent intersubject variability of results.