This dissertation presents an interdisciplinary investigation into the construction of gender-transgressive supernatural voices in Gothic media, drawing on works in queer and feminist theory, voice studies, and performance studies. Spanning two centuries and case studies including Gian Carlo Menotti’s The Medium, art and popular song including Franz Schubert’s “Erlkönig” and Kate Bush’s “Leave It Open,” literary works by Charles Brockden Brown and Vernon Lee, and the Spiritualist séances of Louisa Ann Meurig Morris and Jesse Shepard, I argue that these gender-transgressive voices offer a striking alternative theorization of the “disembodied voice” that, in direct contrast to techno-determinist narratives, is created in performance through unsettling the voice-body-gender relationship. I locate the origins of the connection between gender-transgressive and supernatural or disembodied voices in the early nineteenth century, where rapidly changing ideas about of gender and the body collided with a parallel retheorization of voice as both an important locus for understanding social difference and a site of identity formation. The Gothic became an important mode to explore and destabilize the relationship between voice, body, and gender, particularly for voices that did not conform to increasingly rigid gender expectations; high male and low female voices are consistently used to mark alterity in Gothic media across genres, as are other queer-coded vocal acts. This context sets the stage for what I term performed disembodiment; moments in which a voice is understood as being disembodied, despite the visible presence of the vocalizer. My work argues that some forms disembodiment can be produced not by making a performer’s body absent, but precisely through marking the body’s presence and setting the performing body at odds with the voice through gender-transgressive techniques. One of the primary methods of effecting this performed disembodiment is through “cross-gender vocalization,” wherein pitch, timbre, and articulation are manipulated resulting in, for example, female bodies that appear to produce “male” voices. My dissertation thus argues that “disembodiment” can be produced not only via technologies but through contextual strategies of performance, involving both performers and audiences in the creation of the disembodied voice