Is a joke funnier if it is told by a man? A novel paradigm using AI-generated voices

The stereotype that “men are funnier than women” is culturally pervasive. In a large preregistered online study (n=966), participants rated 12 jokes delivered by a male or female AI-generated voice. To control for differences in delivery, cadence, and intonation, we used a speech-to-speech transformation to transform jokes spoken by a female into a male version and vice versa. Using a Bayesian analysis, we found strong evidence that the apparent gender of the speaker does not affect the perceived funniness of the joke (BF=0.023). We found no interaction between the original and apparent voice gender on masculinity ratings, indicating that vocal gender cues were effectively eliminated when voices were transformed to the opposite gender, reinforcing the merit of this gender manipulation method. Further, men had significantly higher agreement that being funny is a masculine trait than women (MD=0.73, padj<.001), and participants with more male than female friends had significantly higher agreement with this belief than those with equal numbers of male and female friends (MD=0.68, padj<.001) or more female than male friends (MD=0.91, padj<.001). In all, apparent vocal gender does not appear to influence humour evaluations directly, but exploratory findings suggest the apparent vocal gender and delivery may shape broader impressions of speakers (such as perceived confidence and charisma). And individual differences such as gender and proportion of male friends may shape gendered attitudes and expectations toward humour.