Sample Orchestrator : gestion par le contenu d'échantillons sonores

The main advances of the R&D Sample Orchestrator project are presented, aiming at the development of innovative functions for the manipulation of sound samples. These features rely on studies on sound description, i.e. the formalization of relevant data structures for characterizing the sounds’ content and organization. This work was applied to automatic sound indexing and to the development of new applications for musical creation interactive corpus-based synthesis and computer-aided orchestration. The project also included an important part on high-quality sound processing, through several enhancements of the phase vocoder model – processing by sinusoidal model in the spectral domain and automatic computation of the analysis parameters. MOTS-CLÉS : musique, son, échantillons sonores, synthèse et traitement sonore, indexation sonore, recherche d’informations musicales, cognition auditive, bases de données multimédia, apprentissage automatique, modèles de signaux, vocodeur de phase, transformée de Fourier à court terme, modèle sinusoïdal, synthèse concaténative, synthèse par corpus, composition assistée par ordinateur, orchestration, algorithmes génétiques, optimisation multicritères, interfaces homme-machine.