Generating 4pi reverberation using virtual sound sources detected from wave-based simulated responses

Advances in Virtual Reality technology have increased demand for acoustical technology which can interactively generate 3-D sound fields in virtual environments. For reproducing 3-D sound fields, specifically generating 4pi reverberation of a target space, two main approaches exist: geometrical acoustics methods like a mirror source method, and wave-based acoustics methods such as FDTD, FEM, and BEM. Geometrical methods offer calculation efficiency and interactivity but lower accuracy, whereas wave-based methods provide accuracy at high-computational cost unsuitable for real-time use. To overcome this, we propose a hybrid method combining the accuracy of wave-based methods with the efficiency of geometrical methods. Our approach performs wave-based acoustic simulation of a virtual space as a pre-computing step. From this, virtual source information is extracted and used to generate 4pi reverberation interactively through the manner on geometrical acoustics. Unlike traditional methods that interpolate impulse responses at multiple locations, our method updates virtual source information to follow the listener’s movement without re-simulations. Since the reverberation is generated as a Scene-Based Audio signal, it can be decodable to any playback format including binaural 2ch. This presentation introduces the flow of generating the reverberation and generation examples in virtual spaces. Comparisons with real-space measurements are also shown.