Individuals with autism spectrum disorder (ASD) present with core deficits in social interaction and communication. The goal of the current study is to decipher the neural mechanisms underlying impaired speech and language processing in autism. Using a linear neural encoding framework, we analyzed the temporal response function (TRF) to continuous audiovisual speech stimulation at different levels of the speech processing hierarchy, ranging from acoustic to phonetic, to understand the stages at which information processing breaks down in individuals with ASD. We recorded high-density electrophysiology (EEG) in children with a diagnosis of ASD (9–22 years, mean ± SD = 13.7 ± 3.44, n = 25) and an IQ-, sex-, and age-matched control group of typically developing (TD) children (8–18 years, mean ± SD = 12.93 ± 3.00, n = 18). Videos of an actor reciting a children’s book were generated, and randomly interspersed auditory, visual, and audiovisual versions were presented in runs of ~30 seconds while EEG and eye tracking were recorded. To maintain attentional engagement, participants had to respond to the occurrence of target words via a button press. Preliminary results suggest that, at the acoustic level, neural encoding of the speech envelope is reduced systematically as a function of noise level and that this does not differ between groups. In contrast, we expect group differences in the neural encoding of phonetic-level speech sounds, and in the extent to which the addition of visual speech cues improves the neural encoding of such speech units.