A robust adaptive binaural beamformer for hearing devices

Jinjun Xiao, Zhi Quan Luo, Ivo Merks, Tao Zhang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In a hearing aid application, the performance of an adaptive beamformer is sensitive to errors in the acoustic transfer function or noise estimation. The estimation errors can be caused by factors including the head movement of the hearing aid users, non-stationarity of the environment, or imperfect array calibration. With such inaccurate information, the beamformer not only provides less noise reduction, but also causes undesirable speech distortions. To improve the beamformers performance in such conditions, robust beam-forming algorithms have been proposed in the literature. In particular, to minimize the speech distortions caused by imperfect look direction estimation, a robust binaural beam-forming algorithm has been proposed in [1]. This algorithm utilizes constrained optimization to improve the protection of the target speech. The beamforming adaption is based on a low-complexity iteration method called Alternating Direction Method of Multipliers (ADMM). This paper aims for evaluating the proposed binaural beamforming algorithm by comparing its performance with the Minimum Variance Distortionless Response (MVDR) beamformer in various noisy environments. The benefit of the proposed algorithm is demonstrated through both objective and subjective evaluations. In particular, multiple-microphone recordings on a pair of binaural hearing aids on a mannequin are used. Objective perception scores are calculated and compared. In addition, a subjective evaluation of speech intelligibility using normal-hearing listeners is conducted. Both the objective and subjective evaluation results show the robustness of the proposed algorithm.

Original languageEnglish (US)
Title of host publicationConference Record of 51st Asilomar Conference on Signals, Systems and Computers, ACSSC 2017
EditorsMichael B. Matthews
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1885-1889
Number of pages5
ISBN (Electronic)9781538618233
DOIs
StatePublished - Apr 10 2018
Event51st Asilomar Conference on Signals, Systems and Computers, ACSSC 2017 - Pacific Grove, United States
Duration: Oct 29 2017Nov 1 2017

Publication series

NameConference Record of 51st Asilomar Conference on Signals, Systems and Computers, ACSSC 2017
Volume2017-October

Other

Other51st Asilomar Conference on Signals, Systems and Computers, ACSSC 2017
CountryUnited States
CityPacific Grove
Period10/29/1711/1/17

Keywords

  • Microphone array signal processing
  • acoustic beamforming
  • binaural signal processing
  • speech intelligibility

Fingerprint

Dive into the research topics of 'A robust adaptive binaural beamformer for hearing devices'. Together they form a unique fingerprint.

Cite this