How do differences in voice characteristics affect cochlear implant users’ speech intelligibility in cocktail-party scenarios?

Abstract

Speech intelligibility in the presence of multiple competing talkers — the cocktail-party — is quite challenging for cochlear implant (CI) users. Previous literature has indicated abnormalities in CI users’ perception of principal vocal cues (fundamental frequency, F0, and vocal-tract length, VTL), which have been shown to benefit normal hearing (NH) listeners in distinguishing between target and masking speakers in speech-on-speech (SOS) tasks. The hypothesis posited here is that CI users are also expected to benefit from voice differences between target and masking speakers, albeit at a smaller rate compared to NH listeners. Thus, in this study, the benefit in performance in two SOS tasks was measured for both NH (19 participants) and CI users (18 participants) as a function of systematically varying F0 and VTL of the masking talker relative to the target speaker. Results indicate that, contrary to the posited hypothesis and to NH listeners who benefit from differences in F0 and VTL between target and masker, CI users (i) do not benefit from differences in F0, and (ii) demonstrate a slight decrease in performance as the difference in VTL is increased. These data have implications on sound coding strategies that should be developed to address these perceptual deficits in CI users. [Funding: The University Medical Center Groningen (UMCG), Advanced Bionics, Netherlands Organization for Scientific Research (NWO), VICI Grant No. 91817603.]