Heterogeneity in speech-in-noise recognition by CI listeners

Adult cochlear implant (CI) users exhibit a very large range of speech recognition outcomes, and in addition maybe differentially affected by the same level of background noise. We also notice that CI users take varying amounts of time to acquire speech recognition in quiet, and equivalent levels of performance in noise: For example, some CI users achieve levels of performance within one day, or at least within one month of activation equivalent to normal-listener acoustic ‘vocoder’ simulations, while others require months of experience or never achieve open-set speech recognition (James et al., Ear and Hearing 2018)

Optimal electrode position can have some effect on performance. However subject-related factors such as duration of deafness and early-onset or congenital hearing loss also affect performance, both in the short and long term. Most importantly etiologies which may endanger neural survival and/or distort current paths appear to most severely limit speech recognition outcomes.

I will consider what kinds of effect these factors may have on our conception of speech recognition via cochlear implants; on the ‘effective channels’ analogy, and our attempts to improve the performance of CI users in noise via changes in sound processing and coding.