VOICE SYNTHESIS, SIMULATION, AND MEDIATION IN VICTORIAN POETRY

Abstract

This dissertation considers poetry as a print medium for the voice, one participating in the complex media environment ushered in by the nineteenth-century development of communications technology like the telegraph, phonograph, and telephone, and paralleled by the development of voice synthesis technology today. Recent advances in generative artificial intelligence have enabled machines to synthesize and simulate voices—both acoustically and in print—with startling fidelity. I situate voice synthesis as a media practice with origins in the nineteenth century and read Victorian poetry into this media history. Generative AI raises questions about subjectivity, materiality, fidelity, and truth in relation to the voice, and I bring these questions to bear on Victorian poetry. My work examines Robert Browning, William Morris, Amy Levy, and Toru Dutt. These authors demonstrate the way a mediated voice can interrogate, embrace, or problematize the materiality or textuality of their chosen medium: print. The sound of Browning’s poetry, described since the nineteenth century as stuttering or unpronounceable, foregrounds the way hissignature genre—the dramatic monologue—synthesizes voice as a formal effect. William Morris embraced a thoroughgoingly simulative aesthetics, down to the level of the page, creating art out of the mediated condition of history. The work and reception of both Amy Levy and Toru Dutt, demonstrate the way the imagination of an author’s body can impact how readers hear their voice in print, mediated by the material text. I contextualize the work of these writers within a long history of voice synthesis, which I trace from the speaking automata of the late eighteenth-century, through Herrman von Helmholtz’s nineteenth-century synthesizer, Bell Lab’s “Voder” in the twentieth century, and contemporary technology like Neural Text-to-Speech, AI chatbots, and the Project Gutenberg Open Audiobook Collection. The development of voice synthesis technology traces changing conceptions of voice and mediation, as scientific interest moved from producing the voice by simulating its source in the vocal organs to simulating its effects on the ear. Likewise, voice synthesis technologies have long looked to achieve a kind of machinic immediacy by abstracting the voice away from the body in order to conceal the involvement of humans in the production of the machine voice. Examining Victorian poetry in this light helps to situate the challenges we confront today, within a longer legacy of media that transformed how we understand and value history, fidelity, authenticity, and truth.