Exploiting spatial diversity for increasing the robustness of sound source localization systems against reverberation

TL;DR

The paper addresses the challenge of robust sound source localization in reverberant rooms and proposes exploiting spatial diversity by combining SRP-PHAT maps from multiple arrays. It defines and analyzes how SRP maps evolve when arrays are placed at different separations and shows that fusing maps via spatial diversity can mitigate reverberation-induced distortions in the likelihood landscape P(r). Through simulations in rooms with reverberation times up to 2 s and real-office measurements, the authors demonstrate that smaller, spatially separated arrays can outperform a single large aperture under high reverberation and that sum fusion of SRP maps yields more robust localization. The work provides practical design guidelines, indicating that maximizing inter-array separation while maintaining within-array correlation leads to improved SSL robustness with simple map fusion, offering a readily implementable alternative to more complex dereverberation or channel-estimation approaches.

Abstract

Acoustic reverberation is one of the most relevant factors that hampers the localization of a sound source inside a room. To date, several approaches have been proposed to deal with it, but have not always been evaluated under realistic conditions. This paper proposes exploiting spatial diversity as an alternative approach to achieve robustness against reverberation. The theoretical arguments supporting this approach are first presented and later confirmed by means of simulation results and real measurements. Simulations are run for reverberation times up to 2 s, thus providing results with a wider range of validity than in other previous research works. It is concluded that the use of systems consisting of several, sufficiently separated, small arrays leads to the best results in reverberant environments. Some recommendations are given regarding the choice of the array sizes, the separation among them, and the way to combine SRP-PHAT maps obtained from diverse arrays.

Figures (9)

• Figure 1: Comparison of the GCCs for the same pair of microphones and sound source position in anechoic and a reverberant conditions (reverberation time, $RT = 0.8\ \mathrm{s}$) for a microphone separation equal to 0.5 m (left) and 3 m (right).
• Figure 2: Root mean square error in the estimation of the DOA as a function of microphone distance for a sampling frequency equal to 44.1 kHz.
• Figure 3: SRP-PHAT maps generated for a small (left) and a large (right) microphone array. The red points indicate the simulated microphone positions, the filled triangles mark the simulated source position, and the empty triangles show the estimated sound source position. This is a 2D representation at the height of the estimated position. The simulated room has a reverberation time equal to 1.8 s.
• Figure 4: SRP-PHAT maps generated for two different small microphone arrays (left and middle), and the SRP-PHAT map resulting from combining the previous ones (right). Simulation conditions are the same as in Fig. \ref{['fig:srpMaps_distance']}.
• Figure 5: Array topology and position within the simulated room (up), and relative array orientations when two arrays are simulated simultaneously (down).