Factors affecting speech understanding in gated interference: Cochlear implant users and normal-hearing listeners

Previous work [Nelson, Jin, Carney, and Nelson (2003), J. Acoust. Soc. Am 113, 961-968] suggested that cochlear implant users do not benefit from masking release when listening in modulated noise. The previous findings indicated that implant users experience little to no release from masking when identifying sentences in speech-shaped noise, regardless of the modulation frequency applied to the noise. The lack of masking release occurred for all implant subjects who were using three different devices and speech processing strategies. In the present study, possible causes of this reduced masking release in implant listeners were investigated. Normal-hearing listeners, implant users, and normal-hearing listeners presented with a four-band simulation of a cochlear implant were tested for their understanding of sentences in gated noise (1-32 Hz gate frequencies) when the duty cycle of the noise was varied from 25% to 75%. No systematic effect of noise duty cycle on implant and simulation listeners' performance was noted, indicating that the masking caused by gated noise is not only energetic masking. Masking release significantly increased when the number of spectral channels was increased from 4 to 12 for simulation listeners, suggesting that spectral resolution is important for masking release. Listeners were also tested for their understanding of gated sentences (sentences in quiet interrupted by periods of silence ranging from 1 to 32 Hz as a measure of auditory fusion, or the ability to integrate speech across temporal gaps. Implant and simulation listeners had significant difficulty understanding gated sentences at every gate frequency. When the number of spectral channels was increased for simulation listeners, their ability to understand gated sentences improved significantly. Findings suggest that implant listeners' difficulty understanding speech in modulated conditions is related to at least two (possibly related) factors: degraded spectral information and limitations in auditory fusion across temporal gaps.