(Their CHM algorithm had previously shown promising results for pattern extraction from audio recordings.) A number of other presenters likewise covered “automated” approaches to pattern recognition in various genres. Matevž Pesek, Manca Žerovnik, Aleš Leonardis and Matija Marolt applied their Symbolic Compositional Hierarchical Model (SymCHM), an unsupervised learning algorithm designed to extract patterns from symbolic notation, to model similarity among tunes in a Dutch folk song corpus. The tool was developed for comparison of recorded folk melodies, but Savage presented on a newer, and decidedly more practical application of the measure: predicting outcomes of music copyright infringement lawsuits. Patrick Savage, Charles Cronin, Daniel Müllensiefen, and Quentin Atkinson presented a “percent melodic identity” measure which uses an “automatic sequence alignment algorithm” adapted from molecular genetics to quantify melodic similarity between audio files. One of the interesting themes that emerged at this year’s meeting was the issue of transcription, including automated and MIR approaches. The conference provided a congenial atmosphere that was especially notable for bringing together international scholars from more than 20 countries, (1) representing a variety of disciplines, including ethnography, anthropology, cognitive science, computation/music information retrieval, and music theory. The Folk Music Analysis International Workshop met jointly with the AAWM with its own parallel sessions (bringing the total number of presentations up to around 85), and adding a third keynote address (Emmanouil Benetos, Queen Mary University, London).

It hosted over 60 papers, with two keynote addresses (Martin Stokes, King’s College, London John Roeder, University of British Columbia), as well as a few concerts, special workshops, and panels.

The 5th biennial Conference on Analytical Approaches to World Music met in Thessaloniki, Greece, on June 26–29, 2018.

Sonic visualiser source code