A study by Finnish Folk Music Institute head archivist Outi Valo together with Henna Tahvanainen and Tuomas Ylönen investigated whether it is possible to identify and classify stylistic features of music from archival material using deep learning. The object of the research was Kaustinen fiddle playing, inscribed into UNESCO's list of intangible cultural heritage of the humanity, which has stylistic features considered to be characteristic, such as so-called syncopated bowing with strong rhythmical accenting. AI could be used, for example, to help classify archival material. The results of the study are presented in the article ”Feature comparison for classification of Kaustinen fiddle playing style from archived recordings using deep learning”.
Photo: At Kaustinen Folk Music Festival it was tested whether AI could recognize Kaustinen fiddling style on live performance. Photo Lauri Oino.