ESM-Based Source Localization Using the Image Source Method (ISM) with an Offset Microphone Array

Sound source localization is essential for tracking sound origins in structural monitoring, speech recognition and industrial diagnostics. The equivalent source method (ESM) is a near-field acoustic holography (NAH) based sound source localization technique. ESM follows the principle of wave superposition. Conventionally, the sound field from the sources is simulated using Finite element method (FEM) and Boundary element method (BEM) based software to understand the ESM performance. These methods are computationally costly. This study utilized image source method (ISM) simulations to analyze the ESM-based source localization and improve computational performance. Source localization was performed at different Gaussian noise levels using regularization techniques such as pseudo inverse, ridge, lasso regression and iterative reweighted least square (IRLS) to address the ill-posed nature of these problems. Additionally, the microphone array is offset with the equivalent source plane to check the robustness of the method. The results show that source localization is accurate with an offset microphone array using sparsity and non-sparsity-based regularization techniques, even up to 30% Gaussian noise. These findings show that the ISM-based simulations provide crucial insight into analyzing low-frequency source localization.