A Custom Feature Set For Analyzing Historically Informed Baroque Performances

Nolan Eley (1, 2) , Théodora Psychoyou (3) , Catherine Lavandier (2, 4) , Brian F. G. Katz (1)

1 IJLRDA-LAM - Lutheries - Acoustique - Musique
2 ETIS - UMR 8051 - Equipes Traitement de l'Information et Systèmes
3 IReMus - Institut de Recherche en Musicologie
4 Neurocybernétique

A common strategy for analyzing music performances is to extract a high number of low-level features followed by dimensionality reduction. While this can be an effective way of identifying differences in performances, interpreting these differences in a musically meaningful way is difficult. An alternative strategy is to select or design features which are appropriate for the task at hand, however, this must be done carefully and in accordance with musicological principles. In this study, a custom feature set was designed to capture baroque expressiveness in music performances, motivated by principles of historically informed performance practice. This feature set was verified on a small set of annotated performances then applied to a larger group of recordings of historically informed solo baroque performances in two halls, one historically appropriate and one modern. These features were examined to study the effect of room acoustics on the baroque expressiveness of the performances in order to gain insight into the role of acoustics in historically informed performances.

Mots clés: Music performance analysis; chamber music; room acoustics; historically informed performance

Domaines: Son [cs.SD]

Nolan Eley, Théodora Psychoyou, Catherine Lavandier, Brian F. G. Katz. A Custom Feature Set For Analyzing Historically Informed Baroque Performances. ICA 2022 Proceedings, Oct 2022, Gyeongju, South Korea. ⟨hal-03836558⟩