SKM 2023 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
DY: Fachverband Dynamik und Statistische Physik
DY 46: Poster: Machine Learning and Data Analytics
DY 46.1: Poster
Donnerstag, 30. März 2023, 13:00–16:00, P1
Time series analysis of loudness fluctuations in musical performances and psychophysical experiments — •Benjamin Schulz1,2, Corentin Nelias1,2, and Theo Geisel1,2,3 — 1MPI for Dynamics and Self-Organization, Göttingen, Germany — 2Physics Dept., Georg-August University, Göttingen, Germany — 3Bernstein Center for Computational Neuroscience, Göttingen, Germany
Over the last decades, the study of fluctuations in musical time series showed power spectral densities that exhibit a 1/fβ-shape across certain frequency regions, indicating long range correlations. So far time series of pitch, rhythm, or timing were investigated across different musical epochs, composers and styles, showing a variety of β-values between 0 and 2. Whether the fluctuations of musical dynamics, or in other words loudness fluctuations, have similar spectral properties, is an open question, however. We have carried out in-depth studies of manually recorded data sets in different musical settings. A first set results from psychophysical tapping experiments. A second one consists of drum performances recorded in a musical environment. All participating musicians were professionals. The tapping and drumming data consistently show the clear occurrence of a 1/fβ-shape in the power spectral density. Furthermore, the presence of a metronome click in the tapping experiment leads to the strengthening of specific periodic structures in the loudness fluctuations and also seems to have an impact on the coefficient β.