The Sound of a Cochlear Implant: Studies with Single Sided Deaf Listeners

In the past decade, single-sided deaf listeners have received cochlear implants (CIs) for the relief of tinnitus, to improve speech understanding in noise and, most generally, to improve overall quality of life. The data indicate that CIs have been successful in all of these domains. In this talk, I describe a new direction for research with this patient population. Because these patients have one normal hearing ear, we can direct a sentence to the CI ear, and then present sentences that have been modified in frequency content, and other ways, to the normal hearing ear, and ask the patients whether these sentences have the same sound quality as the sentences presented to the CI ear. By using a large library of specially modified signals we can, in the best case, find a modified signal for the normal hearing ear that sounds like the signal presented to the CI ear. First, our results show that CIs do not sound at like either noise or sine vocoders. Second, the sound quality depends on the frequency places stimulated by the most apical electrodes in the array. If the place frequencies are relatively near the input frequencies in the signal (and there is some flexibility), then we find no upshift in voice pitch or formant frequencies. In this case, the dominant perceptual difference between a clean signal and the CI signal is that the CI signal sounds ‘muffled’ to one degree or another. In contrast, when the is some larger distance between the frequency place of stimulation and the frequencies in the input signal, patients match to signals that are upshifted in pitch or in formant frequencies. These voices sound like the little people, or ‘munchkins’, in the American classic movie, “The Wizard of Oz”. Because we can now quantify the voice of an implant, there is a good chance we can improve it.