This presentation highlights an example of integration of state-of-the-art research on audio description into game audio production. We introduce an authoring tool that incorporates content-based classification of environmental sounds to retrieve samples from a user-contributed repository. Given a taxonomy based on ecological acoustics, we show how content-based audio classification, compared to text-only queries, might help sound designers when searching for environmental sounds.

We developed a real-time soundscape generation system with which we can validate the usability of the authoring tool. It allows to create a realistic and interactive soundscape with user-contributed sounds. The soundscape generation consists of a number of graph models, representing virtual acoustic sources, which control the sequencing of events (samples) using concatenative synthesis.