Contribution
Early reflection models for navigable sound field reconstruction
* Presenting author
Abstract:
The growing interest in immersive audio applications, particularly in augmented reality, necessitates an accurate and efficient methodology for characterising sound fields over extended spatial domains. This study addresses early reflection modelling for navigable sound field reproduction, focusing on the perceptual impact of model complexity on externalization. Specifically, we study the predictive performance of three models – monopole, point source cluster, and Spherical Harmonic expansion – for different spatial resolutions based on objective metrics. Moreover, perceptual testing assessed externalisation for two source models – single point source and spherical harmonics expansion up to order 3 – for the cases where the measurement positions were 5 cm and 100 cm apart. Both objective and perceptual results demonstrate that spatial resolution significantly affects externalisation. That is, simpler models perform best when fewer data points are given, and vice versa. This research contributes insights into the relationship between model complexity and perceptual accuracy in spatial audio, with applications in augmented reality and immersive environments. These are preliminary results, and further research is expected to include 6-degrees-of-freedom evaluation.