Unit Test presented our research project 'Not I' at the 12th Conference on Computation, Communication, Aesthetics & X, publishing an article detailing our adversarial engagement with Speech2Face—a machine learning model that reconstructs a speaker's face from voice recordings. Our paper traces the legacy of vocal profiling back to nineteenth-century forensic "vocal portraits" and composite photography, reading modern cross-modal representation learning through the lens of osteobiography— where a physical signal is treated as an autobiographical witness of the body.