Multi-Level Method for Sound Source Location Measurement

This paper presents a novel multi-level search method that enhances the Delay-and-Sum beamformer with an interactive search to identify the location of a sound source in a 3-D space accurately and in a computationally efficient manner. In this method, the delay-and-sum beamformer estimates the sound source location within a cubical search space, with the side length of this cube decreasing in each iteration by a specified reduction factor. In each iteration, a search is conducted within a new cubical space centered on the previously estimated source location. The search process is iteratively refined until estimations reach a state of convergence. Furthermore, our findings across 5 different reduction factors ranging from 0.55 to 0.95 reveal that the choice of reduction factor affects both convergence rate and accuracy; reduction factors closer to 1 provide slower convergence and lower distance error to the actual source, whereas smaller reduction factors provide faster convergence and higher distance error.