Reverberation and early reflections create spurious peaks in acoustic power maps and bias direction-of-arrival (DOA) estimates. At the same time, the direct path and dominant reflections carry useful information about the acoustic environment. This paper investigates joint localization of direct sound and multiple dominant reflected components in single-source reverberant environments using spherical microphone arrays (SMAs). It introduces an iterative spherical-harmonic (SH) diagonal-unloading (DU) beamforming framework for multipath analysis. Microphone signals are encoded in the SH domain and fused into a broadband covariance matrix by frequency smoothing with per-bin trace normalization. Multi-peak localization alternates DU map evaluation and iterative covariance filtering, with detected directions suppressed through a two-sided null-steering update and per-iteration trace renormalization. Performance is assessed through controlled simulations spanning reverberation time, signal-to-noise ratio (SNR), signal-to-diffuse-noise ratio (SDNR), and gain mismatch, together with experiments based on multichannel room impulse responses recorded with a spherical microphone array. Results in terms of root-mean-square error (RMSE) and accuracy rate (AR) show that increasing SH order improves localization of the direct path and strongest reflected components, whereas later iterations become more challenging as weaker components are extracted. The proposed method outperforms the baselines while retaining a computationally attractive structure for SH processing.
Daniele Salvati (Mon,) studied this question.