Individuals consistently identify certain sounds as music but not others (Larrouy-Maestri & Wald-Fuhrmann, pre-registration). In this study, we investigate what grounds a “music” category. To do so, 98 online participants were asked to rate the 90 stimuli selected from various databases on a slider (0: not music – 100: music), which were then clustered into three categories corresponding to music, ambiguous, and non-music stimuli. Each stimulus was described in terms of its acoustic and perceptual characteristics. The acoustic features (n = 252) were extracted with the Essentia toolbox. The perceptual characteristics were rated by participants on ten scales, focusing on aspects such as repetition, intentionality, instrumentality, complexity, melody, rhythm, tempo, pulse, harmony, and timbre. In essence, these perceptual characteristics reflect how listeners interpret the acoustic information. We first reduced both sets of features (acoustic and perceptual) into two meaningful dimensions through principal component analysis and used them as predictors of the three categories (i.e., non-music, ambiguous, and music stimuli) in two separate models. The perceptual model outperformed the acoustic one, supporting the significant role of human interpretation of acoustic characteristics over the acoustic properties themselves in shaping listeners' perception of music.

Larrouy-Maestri, P., & Wald-Fuhrmann, M. (Pre-registrations OSF, 2022-2023). https://osf.io/sb24q, https://osf.io/azkcp, https://osf.io/re7dg