Performance Evaluation After Cochlear Implant And Development And Validation Of Standardized Mandarin Speech Perception Test

Objective:To evaluate the performance of the relatively difficult listening task in cochlear implant (CI) users, which includes complex speech recognition and competing music recognition.1) The speech recognition test was to investigate pediatric CI users’multi-talkers disyllable and sentence recognition performance and how pediatric CI patient demographics influence their speech recognition performance.2) The music recognition task was to explore CIs’melodic contour identification (MCI) performance with competing masker and their ability to use pitch, timbre and timing cues to segregate MCI with masker.3) Also, standardized Mandarin speech perception materials that can be used to evaluate the speech performance of Mandarin-speaking CI users had been developed.Method:1) Open-set recognition of multi-talker (two males and two females) Mandarin Chinese disyllables and sentences were measured in37Mandarin-speaking pediatric CI users. Subjects were grouped according to etiology of deafness and previous acoustic hearing experience. Group1was prelingually deafened, while Group2was postlingually deafened. Multiple linear regression analyses were performed within each group using subject demographics such as age at implantation and age at testing.2) MCI was measured in CI and normal hearing (NH) listeners, with and without a competing masker. The target contours consisted of nine five-note melodic patterns, played by a piano sample. The root note (the lowest note) of the contour was A3(220Hz). The competing masker consisted of a single5-note pattern; the same note was repeated5times. The timing (simultaneous, overlapping and sequential), pitch (A3or A5), and timbre (Piano or Organ) cues were varied between the masker and target contour.3) Ten phonetically balanced Mandarin disyllable and sentence recognition lists were developed. The disyllable recognition materials were validated in8NH subjects listening to unprocessed speech and in10NH subjects listening to a4-channel, sine-wave vocoded acoustic simulation of CI speech processing. The sentence recognition materials were validated in8NH subjects listening to unprocessed speech and in12NH subjects listening to a4-channel, sine-wave vocoded acoustic simulation of CI speech processing. Performance with the disyllable recognition materials was compared to that of the sentence recognition materials for the4-channel, sine-wave vocoded acoustic simulation of CI speech processing.Results:1) Pediatric CI performance was generally quite good. For Group1, mean performance was82.3%correct for disyllables and82.8%correct for sentences. For Group2, mean performance was76.6%correct for disyllables and84.4%correct for sentences. For Group1, multiple linear regression analyses showed that age at implantation predicted disyllable recognition, and that age at implantation and age at testing predicted sentence recognition. For Group2, neither age at implantation nor age at testing predicted disyllable or sentence recognition. Performance was significantly better with the female than with the male talkers (p<0.001).2) Mean NH performance was near perfect across different condition with (96.0%correct) or without the masker (96.6%correct) in music contour identification. With no masker, mean CI performance was61.0%correct (range:37.0%-97.2%correct). With a masker, mean CI performance was47.9%correct (range:8.4%-97.2%correct). NH performance was much better than CI performance (p<0.001). For CI users, the performance without masker was better than that with masker (p=0.025).3) For both disyllable and sentence, there was no significant difference across lists in both unprocessed and4-channel vocoded speech. There was a significant correlation between disyllable and sentence recognition performance.Conclusions:CI users have relatively good speech recognition performance, but still poor music recognition performance, especially the competing music. While some CI subjects were able to use or combine timing, pitch and/or timbre cues, most were not, reflecting poor segregation due to poor spectral resolution. Open-set speech recognition test may be useful in evaluating speech performance with CI users, but still standard test materials which is proper to CI users is needed. The present standardized Mandarin speech perception materials that have been phonetically balanced and validated using spectrally degraded speech may be useful for assessing speech performance of Mandarin-speaking CI users.