In our system, sound sources and sinks are encapsulated as objects having geometry and behavior. Our architecture is based on the VRML2.0 standard. Each sound source has a position and emits sound in a range of an ellipsoid. There are two regions: a core range and an audible range, represented by sound node fields minBack, minFront, maxBack, minFront. The core range determines the full intensity region, while the audible range determines the limit of the audible region. The sound node has a priority for setting the degree of the resource allocation. The direction and location determine orientation and position of the sound node in its local coordinate system. Our application has a visualizer for core range, audible range, intensity, and priority. The source node only specifies direct sound, neglecting room acoustics, but the direct sound allows one to localize the sound. Room acoustic attributes are specified in the Aural Attributes Node. The Sound Sink Node is extension to VRML2.0 specifications. It allows exocentric views and multiple sinks, which allow one to listen at various positions simultaneously within a space [HC96].