The effects of spectral smearing and elevated thresholds on speech-in-noise recognition in simulated electric-acoustic hearing

Abstract

Combined electric-acoustic stimulation (EAS) via hybrid cochlear implants is a relatively new treatment for severe-profound sloping high-frequency hearing loss. Residual acoustic hearing at low frequencies is a vital component of EAS, but it cannot be assumed that residual hearing will be normal; thresholds may be elevated, frequency selectivity may have decreased, or both. In this study, the effects of reduced frequency selectivity and elevated low-frequency thresholds on the intelligibility of speech in noise were simulated. Sentences mixed with a babble masker were processed by an eight-channel vocoder. Residual low-frequency hearing was simulated by replacing the first four channels with the unprocessed acoustic stimulus, with a spectrally smeared stimulus, or with a spectrally smeared and attenuated stimulus. These conditions simulated listening by cochlear implant users with no residual hearing, with normal residual hearing, or with various degrees of impaired residual hearing. Simulating normal residual hearing increased intelligibility, compared to the intelligibility of fully vocoded speech. Low-frequency spectral smearing with filters three or six times broader than normal had little effect on intelligibility. A simulated threshold shift of 60 dB, combined with factor-of-three spectral smearing, significantly decreased intelligibility, but 30 dB of simulated threshold shift did not. These results are consistent with previous results demonstrating the effect of good residual low-frequency hearing on intelligibility. They are also consistent with the use of amplification for individuals who use EAS but have poorer residual hearing.