Comb-filtered speech as a tool to demonstrate difficulties of speech perception and the importance of auditory training in cochlear implant users

Abstract

There are parallels between speech recognition by users of cochlear implants (CIs) and recognition of comb-filtered speech by normal hearing persons. We have found it possible to train inexperienced subjects to recognize spectrally deprived speech and we routinely demonstrate this process to relatives of implanted children to help them understand the complexities of CI auditory perception and training.

We tested the understanding of filtered speech by 38 normally hearing listeners, before and after a 15-minute training period. The original speech recording was comb filtered to retain only three spectral bands of 50 Hz width spaced over the range of 200–6250 Hz. Standard speech audiometry word lists were presented through headphones to one ear at the most comfortable level.

The mean score on the initial (baseline) test was 17% (range 0–37%). Training involved presenting each of the original test words after first informing the subject of the word's identity. The same word lists were then re-presented after the training period and produced a mean score of 57% (37–73%). A third test was then performed using a different word list, producing a mean score of 52% (37–70%).

The variability of results on the first test is analogous to speech recognition differences among newly implanted patients. The high results of the second and third test can be used to show parents that it is possible to learn new skills and improve speech recognition with CIs through training. Parents experiencing the process described here are better able to understand the initially unclear auditory experience of CI users and how this can be improved through training.

Keywords:

• Speech perception
• Spectral deprivation
• Comb filtering
• Auditory training