Dynamic neural oscillations underpin audiovisual gain from visual lip cues in speech noise.

Speech noise substantially impairs speech perception, whereas congruent lip movements provide effective visual support when auditory input is degraded. Using natural continuous Mandarin sentences, the present study recorded behavioral and electroencephalographic (EEG) data under speech noise conditions with different signal-to-noise ratios (SNR) to examine the dynamic neural mechanisms by which visual lip information facilitates speech perception. The results showed that speech noise significantly reduced speech recognition accuracy and weakened δ band activity and functional connectivity in the superior temporal sulcus/posterior superior temporal gyrus (STS/pSTG) and auditory-related regions. Congruent lip movements significantly improved speech recognition in noise and enhanced δ/θ band activity in the STS/pSTG during the early stage after target speech onset. Further analyses showed that, under low-SNR conditions, congruent lip movements increased 11∼15 Hz functional connectivity between the STS/pSTG and motor-related regions. These findings suggest that visual lip information may first facilitate early audiovisual integration by modulating low-frequency neural oscillations and may further recruit sensorimotor networks under difficult listening conditions to provide compensatory support for speech perception in noise.