Contribution of Temporal Fine Structure Cues to Concurrent Vowel Identification and Perception of Zebra Speech

Introduction  The limited access to temporal fine structure (TFS) cues is a reason for reduced speech-in-noise recognition in cochlear implant (CI) users. The CI signal processing schemes like electroacoustic stimulation (EAS) and fine structure processing (FSP) encode TFS in the low frequency whereas theoretical strategies such as frequency amplitude modulation encoder (FAME) encode TFS in all the bands. Objective  The present study compared the effect of simulated CI signal processing schemes that either encode no TFS, TFS information in all bands, or TFS only in low-frequency bands on concurrent vowel identification (CVI) and Zebra speech perception (ZSP). Methods  Temporal fine structure information was systematically manipulated using a 30-band sine-wave (SV) vocoder. The TFS was either absent (SV) or presented in all the bands as frequency modulations simulating the FAME algorithm or only in bands below 525 Hz to simulate EAS. Concurrent vowel identification and ZSP were measured under each condition in 15 adults with normal hearing. Results  The CVI scores did not differ between the 3 schemes (F ^{(2, 28)}  = 0.62, p  = 0.55, η ² _p = 0.04). The effect of encoding TFS was observed for ZSP (F ^{(2, 28)}  = 5.73, p  = 0.008, η ² _p = 0.29). Perception of Zebra speech was significantly better with EAS and FAME than with SV. There was no significant difference in ZSP scores obtained with EAS and FAME ( p  = 1.00) Conclusion  For ZSP, the TFS cues from FAME and EAS resulted in equivalent improvements in performance compared to the SV scheme. The presence or absence of TFS did not affect the CVI scores.