Consultation des notices

The article presents methods for sound search in large effects or instrument sound databases by interactive content-based navigation in a space of descriptors and categories, based on the principle of real-time corpus-based concatenative synthesis. We focus on three algorithms: fast similarity-based search by a kD-Tree in the high-dimensional descriptor space, a mass--spring model with added repulsion for layout, and efficient dimensionality reduction for visualisation by hybrid multi-dimensional scaling based on these. Special attention is given to scalability to very large databases by performance evaluations and measurements. The algorithms are implemented and tested as C-libraries and Max/MSP externals within a prototype sound exploration application.