Cost-effective spatial decomposition method solution for impulse response capture and auralization

Abstract

Spatial decomposition method (SDM) is a method to parameterize a spatial room impulse response into a scalar pressure value and a direction of arrival (DOA) for each time sample. SDM can be therefore be used to simulate acoustics in enclosed spaces and have proven effective when rendering to both headphones and loudspeaker arrays. The aim of this thesis was to design and build a low-cost SDM microphone array using MEMS microphones and investigate how different number of microphones affect auralization performance. All impulse response capture, encoding and decoding was implemented with Python, to be published as an open source library. Python modules were implemented to support rendering as both BRIRs and as virtual loudspeaker signals to be used in external auralization software. Due to hardware limitations the SDM array was limited to 6 microphones, although simulations were carried out to compare a 6 microphone array and a 12 microphone array. These were shown not to have any significant perceptual differences as expected from prior research. Listening tests were conducted where participants compared binaural room impulse responses decoded from SDM with a reference dummy head measurement. Two measurement situations were evaluated, clear line of sight and occlusion between source and receiver. The listening tests showed that neither measurement situation performed better than the other, relative to their respective reference and anchor. No clear difference between the ratings of different configurations of SDM encoding were found. All in all, this implementation proved sufficient at encoding and decoding binaural SDM. With additional post processing the audio quality will get even better