Comparing perceptual segmentation in pulsed and unpulsed music with diverse event densities

Segmentation is a fundamental process in music perception, enabling listeners to identify meaningful boundaries within a continuous auditory stream. This process is crucial for perceiving rhythm and pulse, but more broadly, shapes both cognitive and emotional responses. In pulsed music, a clear, regular beat organises the rhythmic structure (and beat subdivisions as pulses are also normally regular in duration), whereas unpulsed music removes this overt timing reference (e.g., music with expressive rubato; non-isochronous or non-metrical music, traditional or new). While prior research has explored music segmentation, it has usually focused on pulse-driven responses. In this study, we examined how participants segmented diverse pulsed and unpulsed music and assessed the influence of event density (excerpts were grouped into sparse vs. dense groups), onset event patterns, metricality, acoustic features, and personal factors. Sixty-five non-musician participants (defined by non-enrolment in music programmes) listened to 26 musical excerpts and tapped to indicate perceived segmentation points. Bayesian autoregressive time-series analysis of inter-onset intervals revealed that in unpulsed music, segmentation is driven primarily by event density. The acoustic features: spectral centroid, spectral flux, and spectral complexity predicted segmentation boundaries, and RMS (acoustic intensity) showed autoregressive influence up to two lags. We also observed distinct differences in predictive factors between excerpts with sparse and dense event onsets. Personal variables (liking, familiarity, and perceived difficulty of tapping) did not account for additional variance in segmentation. Overall, our findings suggest that listeners use different segmentation strategies for music with sparse versus dense events, opting for a form of structural simplification when confronted with passages characterised by high event density. These results underscore the central role of musical structure in guiding event segmentation, and have implications in both musical and non-musical auditory contexts.