Fuzzy Frequencies: Finding Tonal Structures in Audio Recordings of Renaissance Polyphony

In this study, we explore the possibilities of pitch extraction from audio. We collected a set of over 600 recordings of compositions from the Josquin Research Project corpus and used six state-of-the-art multiple pitch extraction methods to create pitch class profiles and pitch profiles. By comparing these to profiles generated from the JRP encodings we established that Helena Cuesta's MultiF0 method produces the best results. Clustering these profiles gives very similar results to clustering symbolic profiles. This implies that, in the absence of sufficiently large corpora of encodings, audio recordings can be used meaningfully to study tonal structures of Renaissance music

Journal Paper

Music from the Josquin Research Project used in the experiment

MedRen 2024 Presentation

Preliminary results of this study have been presented at he International Medieval and Renaissance Music Conference 2024 in Granada, Spain. For more details, please visit the conference schedule.