Sound rendering with early reflections extracted from a measured spatial room impulse response

Abstract

The acoustic properties of early reflections are not trivial to model when one creates plausible sounding virtual acoustic environments. This paper proposes a novel solution to be used with existing room acoustic modeling systems, such as image-source method, where the synthetic filters for reflection surfaces are replaced with captured real reflection responses. The reflections are extracted from a measured spatial impulse response by analyzing the sound pressure level and the direction-of-arrival to detect prominent individual early reflections. These detected reflections are extracted with a suitable window and are applied in room acoustic modeling algorithms as FIR filters. The presented concept is validated with a listening test, suggesting the material filter length must be over 2 ms. However, an all-round filter length could not be determined due to the dependencies on sound signal content and the room. Further research is therefore required to study the found dependencies, to find the optimal extraction parameters and to validate the filters in practical use-cases.