In this paper we study the robustness of a recently proposed Multi-channel Wiener Filter-based speech dereverberation algorithm to errors in the assumed direction of arrival (DoA) of the target speech. Different subsets of microphones of a pair of behind-the-ear hearing aids are used to construct various monaural and binaural configurations of the algorithm. Via a simulation experiment with frontally positioned target it is shown, that when correct DoA is assumed binaural configurations of the algorithm almost double the improvement of PESQ measure over monaural configurations. However, in conditions where the assumed DoA is increasingly incorrect, the performance of the binaural configurations is shown to deteriorate more quickly than that of the monaural configurations. In effect, for large DoA errors it is the simpler, monaural configurations that perform better.
https://www.aes.org/e-lib/browse.cfm?elib=18070
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.
Learn more about the AES E-Library
Start a discussion about this paper!