Implicit and Explicit Voice Training Effects on Speech-on-Speech Perception and Listening Effort

In multitalker listening situations, intelligibility of the target speech improves when spoken by a personally familiar voice or with trained voices. Yet, the literature is inconclusive on which voice training method, implicit or explicit, is most effective. The objective of the present study was to systematically evaluate implicit and explicit training benefits on intelligibility and listening effort in a speech-on-speech listening task.

Of 32 normal hearing participants, half received implicit voice training via a speech intelligibility task, while the other half received explicit voice training via a speaker recognition task. For both training methods, the same stimuli, uttered by the same talkers, and the same number of sentences were used. Following voice training, all participants completed a speech-on-speech listening task. Stimuli consisted of a target and a simultaneously presented masker sentence, presented at -6, 0, and +6 dB target to masker ratios (TMRs), uttered by a trained and an untrained voice. During this task, in addition to speech intelligibility, pupillometry measurements were recorded to assess listening effort.

There was no significant effect of voice training on speech intelligibility, for either implicit or explicit voice training, and no significant difference in intelligibility scores between the two training methods. Nevertheless, increasing the TMR significantly improved speech intelligibility. A significant interaction between voice training method and TMR was shown, and post hoc analysis revealed better speech-on-speech listening performance after implicit than explicit voice training at +6 dB TMR. Time-based analysis of the pupillometry data showed smaller pupil dilation responses (less effort) for both implicit and explicit voice training groups, in some TMR conditions. For trained voices only, pupil dilations were significantly smaller for explicitly trained voices than implicitly trained voices in the -6 and +6 dB TMR conditions.

Results from the present study demonstrated that both implicit and explicit voice training might lead to a reduction in listening effort, even in the absence of a speech intelligibility benefit. Overall, listening to voices that were explicitly trained seemed to be less effortful than listening to implicitly trained voices. Nevertheless, implicit voice training might be more beneficial in improving speech intelligibility performance for more favorable TMRs. Therefore, both implicit and explicit voice training might provide differential benefits for speech-on-speech perception. Nevertheless, voice training effects reported in the present study can reflect a combination of stimulus-specific perceptual learning and voice familiarity, as the same stimuli were used during training and testing, with generalization to novel utterances left for future studies.