ERASURE is an audiovisual piece that explores the entangled relationship between voice, identity, and technology. Drawing inspiration from the language and format of music videos, the work blends aesthetic experimentation with critical inquiry.
ERASURE integrates generative AI tools into the creative process, pushing the boundaries of how music and voice can be composed. Through techniques such as AI voice model training and voice cloning, ERASURE merges human and digital voices to reveal how artificial systems encode and reproduce social biases. In most of the piece, the human vocalists never actually spoke the words being heard, yet the AI replicates their voices so precisely that it becomes nearly impossible to discern the difference.
By training AI on a group of queer and trans vocalists, the piece celebrates the fluidity and multiplicity of identity while exposing the ethical and technical tensions surrounding representation, authorship, agency, and deepfake technologies. It asks: Who gets to be represented? Who controls a voice? What happens when technology can replicate or alter someone’s identity? By foregrounding the political and poetic dimensions of the voice, the work exposes both the promise and the peril of AI’s attempt to emulate what is most human.