A seismologist’s approach to the analysis of traditional Georgian vocal music
Events
Speaker: Frank Scherbaum, University of Potsdam/Germany
Seismologists sometimes like to refer to their research field as "Musicology of the Earth," a metaphor that holds a surprising amount of truth. The physical mechanisms of "sound" production in the Earth and in musical instruments – including the human voice – are often quite similar, as are the computer tools that can be used to analyze audio recordings and seismic recordings today. In this talk, I will use the similarity between musicology and seismology differently, describing how a seismological perspective on the recording of singing voices could be used to solve the source-separation problem for multiple singing voices in a natural singing environment. This source-separation problem for ethnomusicological field recordings is more than a century old.
Recordings of skin/muscle vibrations close to the larynx, combined with conventional audio recordings (e.g., high-quality headset microphones), turned out to be very useful for subsequent computational analysis, including pitch tracking, tuning analysis, analysis of voice interaction, as well as documentation and archiving purposes.
This idea was systematically applied to the investigation of traditional Georgian vocal music, leading to the international research project "Computational Ethnomusicology of Traditional Georgian Vocal Music (GVM)," conducted at the German universities of Erlangen and Potsdam between 2018 and 2022. The project resulted in the generation of new multi-media datasets of traditional Georgian vocal music, the development of new visualisation techniques for non-Western music, and fascinating insights into the music from Upper Svaneti.
Speaker biography
Frank Scherbaum is professor emeritus of geophysics at the University of Potsdam, Germany. After a career in seismology, he turned in 2011 to the computational study of traditional Georgian vocal music, pioneering the use of larynx-microphone field recordings and signal-processing methods to analyse polyphonic singing voice by voice. From 2018 to 2022, together with Meinard Müller (Friedrich-Alexander-Universität Erlangen-Nürnberg), he led the DFG-funded project "Computational Analysis of Traditional Georgian Vocal Music" (GVM), which produced the curated Erkomaishvili dataset and a new corpus of multichannel field recordings from Svaneti, Guria, Racha, and Tbilisi — the basis of the two studies presented in this talk. In 2018 he founded the SeismoSoundScape-Lab (https://www.uni-potsdam.de/en/soundscapelab#c976942), a virtual laboratory for transdisciplinary work at the intersection of acoustics, signal processing, and musicology. In recognition of his contributions to the study of traditional Georgian vocal music, he was awarded an honorary doctorate by the Vano Sarajishvili Tbilisi State Conservatoire in 2022. An anecdotal account of his career path from seismologist to ethnomusicologist can be found at: https://www.uni-potsdam.de/en/soundscapelab/about-seismosoundscape-lab/people/frank-scherbaum/my-personal-travelogue
2026