Why can’t Siri sing? Cultural narratives that constrain female singing voices in AI

Abstract

Despite the remarkable technological accomplishments in the areas of speaking and singing in AI research, this paper builds upon James W. Carey’s premise that technology is a totem for culture (1990), arguing that YouTube listeners sometimes experience highly charged emotional responses, such as anxiety, pleasure, and musical chills, when listening to the singing voices of Siri, Alexa, and Sophia the robot—responses that reveal an underlying cultural narrative that female singing voices (even robotic ones) are simultaneously fascinating and frightening. This narrative supports Kate Manne’s definition of misogyny as an enforcement strategy (2018), which is operative in several ways female computer voices are controlled and contained. Examples of policing mechanisms inherent in creating and responding to female-gendered computer voices include (1) maintaining age-old stereotypes about singing women, (2) programming virtual assistants to be servile, and (3) encouraging ridicule in YouTube videos to downplay the fear that has been associated with female singing voices since the dawn of Western antiquity. Thus, while research in AI continues to develop technologically, examining underlying cultural narratives about women’s singing voices may be at least as important as developing sophisticated algorithms in shaping future directions for creating female-sounding vocality in AI.