The effects of within- and across-ear temporal misalignment between acoustic and simulated electric signals on speech-in-noise recognition.

Abstract

This study investigated the effects of within- and across-ear temporal misalignment between acoustic and simulated electric signals on speech-in-noise recognition. Sentence recognition was measured in speech-spectrum noise (SSN) and amplitude-modulated noise (AMN) among two groups of normal-hearing listeners. Signals were low-pass filtered and five-channel noise vocoded, presented dichotically for bimodal and monaurally for electric-acoustic stimulation (EAS) simulation. Temporal misalignment was implemented by introducing delays between simulated electric and acoustic signals (0, ±5, ± 10, ±15, ±30 ms), with positive delays (group 1) simulating the electric-leading conditions and vice versa (group 2). A three-way interaction indicated that delay effects depended jointly on the listening condition and noise type. Under AMN, the slope of delay was significantly steeper in the EAS condition compared to the bimodal condition. In the EAS condition, the slope of delay was significantly steeper under AMN than under SSN. In addition, the two-way interaction between the lead-lag relationship and delay indicated that the slope of delay was steeper when the simulated electric signal lagged the acoustic signal. These findings suggest that compensating for temporal misalignment between acoustic and electric signals may help enhance speech recognition, particularly under conditions where the impact of delay is greater.