Memory-Reduced DOA Estimators While Using Cyclic-Symmetrical Array

Home-assistant devices usually use microphone arrays for voice capture and enhancement via beamforming. A key parameter for beamforming is the direction of arrival (DOA) of the relevant speaker. The main disadvantage of DOA estimation algorithms is high memory usage due to the need for a steering vector to be prepared and saved for each candidate DOA. Frequently, a cyclic-symmetrical array is used within these devices. In this paper, a technique for reducing the memory usage for such arrays is shown. Moreover, a technique for blind estimation of the fundamental angle is described. Finally, as an experimental example, a cyclic-symmetrical array with 17 microphones is used for DOA estimation using the generalized cross correlation (GCC) technique and the proposed memory reduction technique is demonstrated.