Poetry Performance and Pitch Tracking: Tools for Sound Studies


ACLS Digital Innovation Fellowships




When we say that poet X reads with a neutral or expressive tone, what do we mean? This project adapts and improves pitch-tracking tools—commonly used by linguists to analyze the fundamental frequency of the human voice and intonation patterns—in order to refine methodologies and terminology used by humanities scholars concerned with sound and performance. The first phase will develop a simple, user-friendly interface for pitch-tracking, drawing on two open-source software programs, Praat and ARLO (Adaptive Recognition with Layered Optimization). These tools will then enable new empirical research on recordings of individual performances, as well as distant listening projects (analogous to distant reading), using machine learning on the “big data” of audio archives to study trends in performance and oratory. The project, which draws on the expertise of linguists, machine learning scientists, and media artists and designers, is hosted by the ModLab at the University of California, Davis.