Sonic-Social Genre: Towards Multimodal Computational Music Genre Modelling
Georgina Born, Department of Anthropology and Institute of Advanced Studies, UCL
Owen Green, Postdoctoral Fellow, Max Planck Institute for Empirical Aesthetics, Frankfurt, and Department of Anthropology, UCL
Bob L. T. Sturm, KTH Royal Institute of Technology, Stockholm
Melanie Wald-Fuhrmann, Max Planck Institute for Empirical Aesthetics, Frankfurt
A core topic of music informatics is music ‘similarity’, which is often approached by the proxy task of music genre classification. This proxy task however lacks an adequate theory of genre, as well as meaningful experimental methods for evaluating music listening algorithms. This study addresses these problems through innovative collaborations between music informatics engineers, musicologists, and music anthropologists and sociologists. The aim is to examine how the nature of genre is and is not compatible with computational methods by modelling distinctive combinations of acoustic data and key mediations: eg acoustic, notated and performed, and acoustic augmented by social and cultural dimensions of genre. A central challenge is to find innovative ways to translate the ‘social’ and ‘cultural’ computationally while, in contrast to previous approaches, not rendering them ‘external’ to computation and resisting their reduction to quantification. The study intends to create methodologies that can more reliably identify the successes and failures of music listening systems, with potentially foundational implications for other areas of computational modelling that act on and within socal and cultural domains.
The project centres on sustained interdisciplinary workshops between CS and SSH that will enable and pursue mutual translation between disciplines, with the aim of incorporating new challenges and perspectives from musicology, anthropology and sociology of music, and science and technology studies into music informatics.