E. Hadad, S. Doclo, and S. Gannot, “Binaural LCMV beamformer and its performance analysis,” IEEE Tran. on Audio, Sp., and Lang. Proc., Aug. 2015, accepted for publication.
We have used the database described in . The experimental setup consists of two hearing aid devices, each with three microphones mounted on an artificial head in a cafeteria with a reverberation time of approximately 1.25Sec.
We report the results for various acoustic scenarios, where (θ, d) is a descriptor for a directional source at angle θ and distance d from an artificial head (A-E refer to various source positions in the cafeteria as described in ).
The directional sources are synthesized by convolving clean speech and noise signals with measured BTE-IRs. Babble noise, originating from multiple simultaneous conversations constitutes the diffuse sound field.
η denotes the interference scaling parameter, which controls the amount of interference reduction
The input target to interferer ratio (SIR) and the input target to background noise (SNR) is set to 6dB and 14dB, respectively.