How the human voice must be understood, framed, and transformed to fit the conceptual formats required for artificial intelligence.
What does it mean to know a voice, and how is that knowledge troubled through attempts to measure, analyze, and generate it? In Machine Voices, Edward Kang explores the sonic beliefs and practices that orbit voice AI technologies. He resists the paralysis of the increasingly ambiguous term “artificial intelligence” and examines it as a cultural hypothesis of human capability. Similar to how computer vision models inherit a limited belief of “how humans see,” or how LLMs employ a narrow proposition about “what language is,” voice AI systems invoke a thesis of how humans listen and speak.
Kang offers the “machine voice” as a critical frame for understanding how the human voice is made commensurate with the mechanical formats required for artificial intelligence. In so doing, Kang not only presents new analytical frames to understand voice AI technologies and the economic, social, and cultural forces underlying their construction, but also pushes the boundaries of voice and listening in ways with which we have not yet contended.