Sexualised synthetic personas encode and amplify gendered power asymmetries through voice.
Published in Interspeech, 2026
Examining sexualised AI-generated English-speaking voices in a popular commercial platform from a Feminist HCI perspective, we find qualitative and quantitative differences in a diverse group of listeners’ reactions to male and female-coded voices, showing how social biases are reinforced through the voices’ conformity to a clichéd, heteronormative vocal attractiveness schema.
Recommended citation: Alice Ross, Ariadna Sanchez, Elin Kanhov, Catherine Lai, and Éva Székely. 2026. Sexualised synthetic personas encode and amplify gendered power asymmetries through voice. (accepted) Interspeech 2026
Download Paper
