The study of spoken language requires controlling multiple aspects of the speech signal. Here we report a procedure to create a sarcastic version of sincerely spoken audio stimuli by changing aspects of prosody relevant to sarcasm (pitch, pace, and loudness) while controlling all other acoustic differences. Two rating experiments validated the efficacy of this procedure for spoken conversations ("Maybe they are more delicate than you realized") and descriptions ("Angie thanked John for doing such a great job"; emphasis indicates manipulation).