Twenty-one Hebrew speakers listened to speech excerpts of 27 Japanese cartoon voices in the random-spliced and non-manipulated conditions and rated their impressions of physical and personality traits, emotional states, and vocal characteristics on 7-point scales. The correspondence of ratings between the two manipulation conditions was examined by calculating Pearson’s correlations for individual participants, and for the mean ratings across participants. Cronbach’s alpha was also calculated to assess inter-rater reliability. Possibilities of systematic biases introduced by the random-splicing technique are discussed.