scholarly journals Deep learning-based direction-of-arrival estimation for multiple speech sources using a small scale array

2021 ◽  
Vol 149 (6) ◽  
pp. 3841-3850
Author(s):  
Min Zhang ◽  
Xiang Pan ◽  
Yining Shen ◽  
Jianjun Qiu
2021 ◽  
Author(s):  
Suat Yetiş ◽  
Özgür TAMER

Abstract In this work, a combined direction-of-arrival (DoA) estimation method for a four element square array is presented. A four element array is a very small planar array and estimation performance with ordinary DoA estimation techniques is considerably low. Main goal in this work is to improve the estimation performance of such an array by estimating the DoA with different geometrical configurations of the array elements and combine the results to evaluate the resulting DoA estimation. The geometrical structures employed are based on the circular, L shaped and linear configurations of the array elements. Performances of the geometries are evaluated and a combining filter based on the weighted results of the geometries is generated. The weighted results evaluated at the output of the filter are superior when compared with individual results of each of the geometries.


Sign in / Sign up

Export Citation Format

Share Document