Distance Mapping for Corpus-Based Concatenative Synthesis

Diemo Schwarz (1)

1 Equipe Interactions musicales temps-réel

In the most common approach to corpus-based concatenative synthesis, the unit selection takes places as a content-based similarity match based on a weighted Euclidean distance between the audio descriptors of the database units, and the synthesis target. While the simplicity of this method explains the relative success of CBCS for interactive descriptor-based granular synthesis — especially when combined with a graphical interface — and audio mosaicing, and still allows to express categorical matches, certain desirable constraints can not be formulated, such as disallowing repetition of units, matching a disjunction of descriptor ranges, or asymmetric distances. We therefore propose a new method of mapping the individual signed descriptor distances by a warping function that can express these criteria, while still being amenable to efficient multi-dimensional search indices like the kD-tree, for which we define the preconditions and cases of applicability.

Mots clés: Constraints; concatenative synthesis; unit selection; databases; content-based retrieval; audio mosaicing; corpus-based synthesis; audio descriptors

Mots clés: Informatique musicale

Domaines: Son [cs.SD]; Interface homme-machine [cs.HC]; Musique, musicologie et arts de la scène; Traitement du signal et de l'image [eess.SP]; Apprentissage [cs.LG]; Intelligence artificielle [cs.AI]; Ingénierie assistée par ordinateur; Multimédia [cs.MM]; Vision par ordinateur et reconnaissance de formes [cs.CV]; Autre [cs.OH]; Traitement du signal et de l'image [eess.SP]

Diemo Schwarz. Distance Mapping for Corpus-Based Concatenative Synthesis. Sound and Music Computing (SMC), Jul 2011, Padova, Italy. pp.1-1. ⟨hal-01161294⟩