This paper discusses an experiment in which 32 subjects listened to random-spliced excerpts from the speech of 27 cartoon characters and rated their impressions of age, gender, physical and personality traits, emotional states, and vocal characteristics. Statistical analyses were performed in order to examine the consistency of participants' trait ratings and the relationship between the auditory characteristics of the voices and subjects' trait ratings of the speakers. The results suggest that random splicing can be a useful method of examining the effects of voice quality on impression formation.