Improvements in acoustic source’s position and orientation estimation by a microphone array network

In this work we present improvements in our research on position and orientation estimation of an acoustic source by a microphone array network. Here, at the input of an artificial neural network (ANN), combinations of energy related features, time delay of arrival (TDOA) of microphone pairs, and source position candidates derived for every array of the array network are directly used in a mapping process to estimate more accurately the position and the orientation of the source. This approach has the advantage of exploring the spatial information in two-/three dimensional (2D/3D) spaces compared to one-dimensional information generated by a distance information between an array and its corresponding position candidate used in our previous research. TDOAs are estimated by a robust GCC-PHAT (Generalized cross-correlation with phase transform) function, and position candidates are estimated by two distinct methods; a TDOA-based and the SRP-PHAT (Steered Response Power Phase Transform) estimation methods.