Numerous approaches have been taken to address the problem of generating navigable virtual models for multi-volume acoustic spaces. The general practice for creating empirically informed interactive models of multi-volume acoustic spaces, as embodied by the Spatially Oriented Format for Acoustics, is to discretely sample emitter-receiver pair positions. For a user to then navigate between these discrete positions involves cross-fading, blending, or otherwise perceptually interpolating between corresponding zones. This paper outlines a new approach which instead involves the continuous three-dimensional sampling of acoustic spaces, much as is done with 3D visual spaces in photogrammetry. To achieve this result, a first-of-its-kind consolidated ambisonic impulse response capturing apparatus has been designed and built. This apparatus combines a 3rd-order ambisonic microphone array with a 2nd-order ambisonic loudspeaker array and is designed to be moved through a space with maximal ease. AD/DA conversion, playback, and recording are all handled on a central compute platform. In parallel, a software workflow has been developed which can be implemented in Unreal Engine, as well as other game engines. To solve general issues of spatial audio in game engines, a custom encoding and decoding framework has been implemented. Then, to map the continuous ambisonic impulse response onto a virtual space, a spline mirroring the sampling path is drawn through the space. On the DSP side, an impulse response is extracted from any arbitrary point along the spline by way of the Common-slope Model for coupled spaces. Future work for better addressing early reflections and minimizing the theoretical intermediary of the Common-slope Model is discussed. Additionally, a special use case for visualizing acoustic energy in architectural acoustics is explored.
