This paper presents the results of a Cantonese lexical tone normalization experiment. Two non-linguistic (ie, music and nonspeech) context conditions were used in addition to a speech context condition to elicit listeners’ normalization of level tones following contexts with fundamental frequency alterations. Participants showed clear tone normalization in the speech context condition. Participants’ normalization performances in the non-linguistic conditions were not significant but comparable to each other. The findings indicated that the non-linguistic music context is not sufficient to support lexical tone normalization.