The impulse response of a room describes the acoustic transmission path between a given source and receiver. One way to acquired an impulse response is by using simulations of a virtual environment. With auralization techniques, such an impulse response can be employed to listen to the virtual room. In a following step, these simulations can be included into interactive virtual reality environments. For this, the simulations must be performed under real-time conditions, since the auralization has to be updated according to the user-input. For simple acoustic scenes real-time constraints can be met. However, if the complexity of the room increases, the allowed simulation run-time for real-time update rates is often exceeded.

To address this issue, the presented work proposes a scheduling scheme which decreases the amount of necessary simulations by detecting inaudible simulation inquires from the interactive virtual environment. This detection is done by adjustable prediction models that apply technical and perceptual thresholds. For example, only if the movement of source and/or receiver is predicted to cause a significant audible change, compared to the previous simulation result, the simulation is carried out.

The implementation and performance of the proposed modular scheduling system and audibility prediction filter algorithms are presented. First, possible thresholds for the filters are discussed. Then, the real-time performance of the system with different configurations is evaluated by a real-time benchmark. Based on these results, a configuration of the system is derived, that can be applied for various rooms of different complexity.