Vocal imitation in English-speaking autistic individuals has been shown to be atypical. Speaking a tone language such as Mandarin facilitates vocal imitation skills among non-autistic individuals, yet no studies have examined whether this effect holds for autistic individuals. To address this question, we compared vocal imitation of speech and song between 33 autistic Mandarin speakers and 30 age-matched non-autistic peers. Participants were recorded while imitating 40 speech and song stimuli with varying pitch and duration patterns. Acoustic analyses showed that autistic participants imitated relative pitch (but not absolute pitch) less accurately than non-autistic participants for speech, whereas for song the two groups performed comparably on both absolute and relative pitch matching. Regarding duration matching, autistic participants imitated relative duration (inter-onset interval between consecutive notes/syllables) less accurately than non-autistic individuals for both speech and song, while their lower performance on absolute duration matching of the notes/syllables was presented only in the song condition. These findings indicate that experience with tone languages does not mitigate the challenges autistic individuals face in imitating speech and song, highlighting the importance of considering the domains and features of investigation and individual differences in cognitive abilities and language backgrounds when examining imitation in autism.