목적: Audiovisual (AV) integration plays a fundamental role in speech perception, allowing individuals to extract meaningful linguistic information even in challenging environments. The McGurk effect, a perceptual phenomenon in which an incongruent pairing of auditory and visual speech stimuli leads to an illusory percept, serves as a robust paradigm to examine multisensory integration mechanisms. Previous studies have suggested that musical training enhances auditory processing, sensorimotor coordination, and predictive coding, yet its specific impact on AV speech perception remains underexplored. Given that musicians are frequently exposed to complex auditory-visual interactions, it is plausible that their long-term training influences susceptibility to the McGurk illusion and its underlying neural correlates. The present study aims to investigate how long-term musical training modulates behavioral responses and neural activity during AV speech integration, with a particular focus on alpha power dynamics as a neural marker of predictive processing. 방법:To examine the effect of musical training on AV integration, we recruited thirteen musicians and eleven age-matched non-musicians. Participants performed a McGurk task while undergoing electroencephalography (EEG) recording. The task consisted of three experimental conditions: an audio-only (A-only) condition, where participants heard the syllable /pa/ without visual input; a congruent (ApaVpa) condition, in which auditory /pa/ was synchronized with visual /pa/; and an incongruent (ApaVka) condition, in which auditory /pa/ was paired with visual /ka/, inducing the McGurk illusion. Each participant completed a minimum of 100 trials per stimulus type. EEG data were analyzed using time-frequency analysis to examine alpha power modulations associated with AV integration. To further localize the neural generators of these effects, we employed Dynamic Imaging of Coherent Sources (DICS) analysis. 결과:We analyzed five experimental conditions, each associated with distinct response patterns: Aonly-A, Aonly-F, Congruent, Incongruent- A, and Incongruent-F. Behavioral results revealed a significant condition effect but no overall group effect. In the time-frequency analysis, alpha activity exhibited the most prominent changes during the stimulus period, leading us to focus on alpha activity using DICS source analysis. Consistent with the behavioral findings, no significant group effect was observed in the neural data; however, a within-group effect emerged exclusively in the musician group. Additionally, we examined the unexpectedly high rate of illusion responses in the A-only condition. Notably, the likelihood of perceiving an illusion in this condition was influenced by the nature of the preceding trial (T-1), suggesting that prior perceptual experience played a critical role in shaping subsequent responses. In contrast, non-musicians exhibited more stable response patterns across trials, indicating a reduced influence of prior stimuli on their perceptual judgments. 결론:Our findings suggest that long-term musical training enhances audiovisual speech perception by increasing sensitivity to prior perceptual experience, as evidenced by a within-group effect observed only in musicians. Alpha activity, which showed the most significant modulation during the sound onset period, was primarily altered in musicians, indicating that musical expertise strengthens predictive coding mechanisms and neural plasticity in multisensory speech processing.