Toward a Hierarchical Model for Audio Scene Analysis and Archiving, by Manuel Moussallam

Wednesday, June the 23rd 2010, 1:30pm
Télécom ParisTech (Dareau)
Vitrine de la Recherche (DA006)

Manuel Moussallam, PhD student

The talk will focus on the problem of efficiently archiving audio scenes. Both the compression and the indexation constraints are to be
taken into consideration to allow efficient storage and fast semantic data access. In an audio context, sparse approximation tools,
particularly decompositions on overcomplete dictionaries, provide great opportunities for simultaneously adressing these constraints, while allowing the automatic construction of a hierarchical model of the data.

Leave a Reply