On Thursday, January 13th, at 15:30h in room 52.321 Jordi Janer will give a seminar on "Soundscape modeling: description and generation using unstructured sound databases"

Abstract: A principal role of audio in virtual environments is the creation of a sound ambiance or soundscape. In the context of a research project at the MTG (www.metaverse1.org), we explored the description and generation of soundscapes around the collaborative sound repository Freesound.org.

Our research aims at simplifying the authoring process, offering at the same time a realistic and interactive soundscape as a result. Graph models, representing virtual acoustic sources, control the sequencing of events (samples) using concatenative synthesis. Samples are retrieved from a user-contributed audio repository, which integrates a content-based search engine and an ecological acoustics taxonomy. A server architecture allows the synthesis engine to support several independent listeners simultaneously. Client applications are responsible for sending position updates, receiving the soundscape as a web stream.