Comparing a BMVDR with an IBM regarding SRT and subjective listening effort
Our study investigated the effects of a binaural minimum-variance distortionless response beamformer (BMVDR) and an ideal binary mask (IBM) on perceived listening effort and speech intelligibility in comparison to an unprocessed condition. Two different spatial listening situations were used by convolving stimuli with head-related impulse responses recorded in a cafeteria. The first condition consisted of two interfering speakers that were situated at +90° and -90° and the target at 0°. In the second condition all speakers were located at 0°. We measured the speech recognition thresholds (SRTs) and categorically scaled subjective listening effort in the same session.The date showed that, in general, collocated situations were rated as more difficult and the SRT was also worse. At 90° the BMVDR had a clear advantage over the unprocessed condition at 90° regarding both SRT and listening effort. In terms of SRT, the IBM way outperformed all other conditions. However, regarding listing effort we found no differences between IBM and BMDR processing at 90°, suggesting that the BMVDR algorithm is a well suited candidate to produce release from masking in moderately reverberant conditions in terms of perceived listening effort.