Table 1.
Pattern Factor Loading, Communalities, Variance, and Factor Correlations Based Upon a Principal Components Analysis (PCA) Factoring with Oblim Oblique Rotations for 16 Audio Feature Variables.
Grey areas indicate the low-level audio descriptors that characterize the respective factor.
Fig 1.
Results of K-means clustering of the stimuli for musical samples.
(A) Results are derived from the five factor scores gained from the low-level audio descriptors (see Table 1). Stimuli are labeled with their Track IDs and plotted in a 2-dimensional representation (Component/Factor 1 and 2) of the four clusters (1 = black, 2 = grey, 3 = green, 4 = pink) presented in Table 2.
Table 2.
Final Four Cluster Centers by Five Factors comprising 19 low-level audio descriptors.
Fig 2.
Distribution of the musical stimuli.
(A) Musical stimuli are drawn from 11 sub-genres of EDM across the 4 clusters derived from the audio descriptors.
Table 3.
Summary Statistics from the MFR-RS Model showing overall significance for all 6 facets (Musical Sample, Rater, Items, Subgenres, Musical Preference and Clusters derived from the Audio Analysis).
Fig 3.
Visual depiction of the cluster calibration in relation to item calibration on the logit scale.
(A) Cluster spread ranged from 0.01 logits to 0.16 logits and demonstrated significant distinction according to one fit item: Inartistic/Artistic (C8).
Table 4.
Calibration of the Item Facet.
Items (C = cognitive, A = affective, P = psychomotor) are ordered by their endorsability. Most difficult item to rate (C9) on top, easiest (C1) on bottom. C, A, and P items are mixed in their ratings and spread across the logit scale.
Table 5.
Calibration of the Cluster Facet.
Spread of the clusters across the logit scale.
Table 6.
Summary of Differential Facet Functioning Statistics (Music Preference Interactions) for Item Exhibiting | Z | > = 2.0.
Showing only selected items, which were significantly overrated by listeners with specific musical preferences.