Fig 1.
Feature values along album segments.
Standardized feature distribution of 51000 albums found by [12]. Given a varying number of tracks, albums were discretized into 3 segments (i.e., beginning, middle, end). Feature values represent the average feature for all tracks pertaining to a given album segment.
Fig 2.
Computer interface used to sequence tracks and rate survey propositions.
Fig 3.
Track frequency by position and set.
PR by position and set, as well as PR across all sets (overall). Shaded area represents the standard deviation for each position.
Fig 4.
Left: density plot of PRsimulated with respect to PRtrue obtained from participant-sequenced albums. Right: PRtrue (dashed lines) in relation to the Empirical Cumulative Distribution obtained from the control condition (solid line), where the y axis indicates the probability of obtaining a value equal to, or greater than a given PR.
Fig 5.
Feature distribution per set and averaged across all sets. Shaded area represents the standard error of the mean for each position and feature.
Fig 6.
Empirical cumulative distributions of track similarities.
Key dissimilarity, tempo dissimilarity and overall feature dissimilarity between pairs of consecutive tracks in the experimental condition (dashed lines), in relation to the Empirical Cumulative Distribution in the control condition. Values in the y-axis represent the probability of obtaining a given value or higher under the current ECDF.
Fig 7.
Distribution of survey responses.
Density plots show the distribution of opinions throughout a continuous scale between completely agree and completely disagree. Bar plots show the proportion of respondents who agreed, disagreed, or were neutral about a given proposition.