On the Acoustic Qualities of Dynamic Pseudo-Binaural Recordings
The motion-tracked binaural (MTB) technique allows the dynamic, pseudobinaural rendering of spatial sound scenes recorded by a circular array of microphones on a rigid sphere. The system provides a multichannel live audio transmission from which a head-related signal with approximated interaural time and level differences can be derived and played via headphones, head tracking, and a corresponding rendering software. The latter is mainly calculating imperceptible interpolation between channel pairs during head movements. This contribution evaluates the potential of this format for the creation of virtual acoustic envi- ronments. Based on the technical realization of a 16-channel MTB array with omnidirectional diffuse field-corrected electret condenser microphone capsules, the plausibility of 8 and 16-channel recordings was tested against a physical sound source. Furthermore, the sound quality of the pseudobinaural rendering was assessed based on different items of the Spatial Audio Quality Inventory (SAQI) compared to a true dynamic binaural reference. The results show that the overall plausibility of the MTB signal with optimal interpolation is close to the reference. Even if there are small differences with respect to tone color and spatial sound source attributes, the degree of externalization and even the perceived source elevation were, despite the absence of pinna cues, well comparable to the true binaural reference.
Published in: Journal of the Audio Engineering Society, 10.17743/jaes.2020.0036, Audio Engineering Society (AES)