Extended high-frequency hearing sensitivity facilitates predictive speech perception

Speech signal decoding relies on both cognitive performance and the ability to perceive the acoustic signal in sufficient detail. In everyday communication, natural speech is often rapid, spatially and spectrally complex, and accompanied by noise. The brain generates predictions to cope with the complex nature of the auditory signal in speech perception. Anticipatory coarticulation is a natural part of the speech signal. Acoustic traces of upcoming speech sounds provide valuable information that can support auditory prediction, especially when the input is limited or degraded. This study analyzed the relation between individual extended high-frequency hearing threshold differences and word recognition performance in an auditory gating task. Using an adapted gating paradigm with fricative onset (/f/ and /s/) words, we investigated listeners' extended high-frequency hearing sensitivity in relation to their ability to use early coarticulatory cues predictively for spoken-word recognition. Listeners with better extended high-frequency thresholds were more accurate in predicting the identity of spoken words, using coarticulatory cues as early as 15 ms for words beginning with /s/. In contrast, /f/, which has lower energy in high-frequency energy bands, did not show this pattern. These findings highlight the perceptual role of extended high-frequency hearing in speech perception in natural speech, where input is often limited and variable, showing that extended high-frequency hearing can facilitate predictive processing by enhancing access to subtle early cues in the speech signal.