Autism is a neurodevelopmental difference characterised by differences in social communication and interaction, influencing the experiences of autistic individuals and families. While autistic individuals often exhibit atypical speech processing, the processing of song appears relatively preserved, showing promise for improving communication. However, the behavioural, neurophysiological, and social mechanisms underpinning speech and song processing in autism remain unclear. This study addresses this gap through three interconnected studies, while also considering individual differences in musical and cognitive abilities. Study 1 examines behavioural mechanisms by testing how autistic individuals perceive and imitate acoustic features along a speech-song continuum. Study 2 develops dynamic Temporal Response Function modelling of EEG data to capture non-stationary neural dynamics and applies it to investigate whether autistic individuals exhibit atypical neural entrainment to speech and song during passive listening. It also evaluates the new tool’s potential for identifying neural markers associated with autism. Extending this to a social context, Study 3 employs hyperscanning to explore inter-brain synchrony between autistic participants and a confederate during spoken versus sung communication. The findings will reveal whether and to what extent song is processed differently than speech in autism, and how individual differences impact the results. This study bridges the gap between behavioural observations and neurophysiological findings, offering a comprehensive picture of speech and song processing in autism. Theoretically, it clarifies the shared mechanisms of music and speech; methodologically, it develops new analysis tools for characterising brain dynamics; practically, it provides neural evidence to optimise music-based interventions for improving social communication in autism.