This dissertation argues that the current preponderance of femme-styled AI voiceassistants (AI VAs) represents part of an evolving communication infrastructure which reifies women’s voices as at once unique and universal in their helpfulness, while also normalizing the invisibility and immateriality of their labor. To trace the evolution of this gendered, sonic infrastructure, I conduct a critical discourse analysis with attention to four key historical moments: the dawn of telephony and the concomitant American “voice culture” obsession; the reign of the Bell Telephone Company, whose women operators were trained to deliver “the voice with a smile”; the increased automation and technological displacement of live operator service; and finally, the modern zeitgeist of popular AI voices. In each era, I examine popular and official discourses surrounding the role of women’s voices in telecommunications labor with special attention to corporate communications from the AT&T Archive & History Center in Warren, NJ. Throughout this analysis, I turn to Susan Leigh Star’s multi-dimensional definition of infrastructure to consider how such discourses act as sites of “encoding and standardizing” which naturalize women’s voices as “ready-to-hand” features of modern communication and information systems. Moreover, I ask: when feminine voices become independent of laboring bodies, where is power? / Media & Communication