Pseudo-intensity vector based sound speed measurement in indoor environments and its application to beamforming calibration

Abstract

This work introduces a passive method for measuring the speed of sound in indoor reverberant environments using speech signals captured by first-order Ambisonic (FOA) microphones. This estimation is a key step in enhancing array signal processing performance, especially in environments where acoustic properties vary due to temperature and humidity changes. By leveraging the pseudo-intensity vector (PIV) derived from FOA signals, we can more accurately estimate the direction-of-arrival (DOA) of a source, independent of the sound speed. Additionally, we present a geometric model to compute the instantaneous sound speed based on the DOA and time delays estimated from signals recorded by two FOA microphones. The study further examines how factors such as source DOA, time delay, and FOA microphone spacing impact the accuracy of the estimate. Both simulations and experiments are conducted to validate the proposed method. Moreover, we demonstrate how instantaneous sound speed estimation can be used for beamforming calibration, facilitating robust beamformer design in time-varying acoustic environments.