Effects of Audiovisual Asynchrony on Speech Intelligibility in Typically Hearing Adults and Cochlear Implant Users

Speech articulators produce auditory cues conveyed by the voice and visual cues that are perceptually integrated to form a highly salient, unified audiovisual signal. How individuals make use of these cues depends on listener- and stimulus-specific factors. Although both typically hearing listeners and cochlear implant (CI) users usually integrate synchronous auditory and visual cues, CI users are more likely to use visual cues to enhance perception. This may be especially true when stimulus characteristics such as temporal alignment of the auditory and visual cues are degraded, affecting intelligibility. Here, we examine the interaction between these listener and stimulus characteristics by quantifying how asynchrony affects speech intelligibility in CI users and typically hearing controls.

Fifteen CI users (aged 59.9 ± 15.5 years) and 31 typically hearing control participants (aged 61.2 ± 15.9 years) were presented with audio-only, visual-only, synchronous audiovisual, and asynchronous audiovisual (100, 200, 300, and 400 ms of temporal offset in both audio and visual leads) sentences in multitalker babble (+9 dB SNR for CI users, 0 dB SNR for controls). Intelligibility was quantified as the proportion of key words correctly reported. Benefits associated with multisensory integration across asynchronies were computed by comparing performance to unisensory conditions.

Intelligibility and multisensory gain were strongest near synchrony for both groups, tapering off as audiovisual asynchrony increased. CI users showed larger decrements in performance with increasing asynchrony than controls. At the largest audio-leading asynchrony, CI users' performance continued to decrease, whereas controls improved relative to smaller asynchronies.

Typically hearing listeners benefit from multisensory integration over a broader range of speech asynchronies than CI users. Interestingly, at extreme asynchronies, CI users appear to be impaired by visual speech cues compared to typically hearing individuals, suggesting that they may attempt to use visual speech cues to inform intelligibility even when they occur outside of the range of asynchronies over which integration typically occurs. These findings align with literature on increased visual bias in CI users and raise questions regarding potential compensatory mechanisms for speech intelligibility.