Poetry lovers, students, and scholars are currently enjoying a moment of unprecedented access to sound recordings of poets reading their work. PennSound, the world’s largest poetry audio archive, offers nearly 60,000 sound recordings and 6,000 hours of material, spanning from 1913 through the present. The existence of such an archive raises the question of what new kinds of research become possible if we think of these collections as datasets. Moving beyond the idea of using sound recordings to elucidate a primary, written text, this talk will demonstrate a number of approaches that consider sound on its own terms. Through machine-listening, artificial intelligence, and new kinds of visualizations of the voice, traditional literary histories can be amended or altogether upended. After a precis of the PennSound archive, we will look digital sound studies approaches at various scales: from tools meant to tarry with the sounds of a single reading through methods for processing large audio corpora. Chris Mustazza is the Co-Director of the PennSound archive, IT director at Penn’s School of Arts & Sciences, and a research fellow at Penn’s Linguistic Data Consortium. He earned his Ph.D. in English and Master’s of Computer and Information Technology at Penn, and his current book project details the birth of the poetry audio archive in linguistic speech labs of the early twentieth century.