PSIND

The PSIND project aims to deepen our understanding of the notion of musical genre via the case study of punk, by objectively disentangling the acoustic substrate of punk sound from its socio-cultural correlates, and revealing the morphologies likely to represent punk sound as closely as possible to the associated musicological labels. The approach combines qualitative methods from SHS that have been well established thanks to a previous ANR (PIND, 2016-2021); an AI-based approach, relevant for testing whether the information contained in sound representations can predict the corpus categories established by musicological discourse; and an analysis-by-synthesis approach that combines sound perception and audio signal processing with a view to characterizing perceptual invariants specific to a category of sounds. To achieve this, PSIND brings together an interdisciplinary consortium (history, musicology, acoustics, computer science), while contributing to the structuring of actions in favor of SAPS (participation of players in the scene). PSIND will make a number of contributions: (1) it will advance the still partial results developed in the field of popular music by proposing an innovative articulation designed to reciprocally nourish the musicological hypotheses elaborated on the basis of SHS tools and the avenues opened up by quantitative analysis; (2) it will validate the use of reverse correlation methods to extract the perceptual invariants associated with a given semantic category; (3) it will enable the development of new methodologies for the automatic calibration of musical synthesizers according to semantic criteria using data-driven approaches ; (4) it will contribute to the preservation of an endangered counter-cultural memory (vulnerability of actors and material supports) by making the studied corpus available to the general public.