Binaural noise reduction combining binaural analysis and psychoacoustically motivated spectral subtraction

In a noisy environment, hearing aid users suffer from decreased intelligibility of speech. To reduce the binaural colored noise in hearing aids, we propose an efficient noise reduction scheme that combines the merits of binaural analysis and perceptually motivated spectral subtraction algorithm. In our approach, the conventional voice activity detector (VAD), widely used in the single channel spectral subtraction, is substituted by a simplified binaural model, which provides more accurate voice activity information even in a low SNR environment. A simple, but effective, perceptually-weighted spectral subtraction algorithm is then applied on the left and right channels respectively, which utilizes the auditory masking threshold to improve the signal quality. Taking into account that most real-world noise is colored with a non-uniform spectrum, the band-specific over-subtraction factors and spectral floors determined by the SNR on each frame are also introduced into the subtraction rules to further reduce the musical noise and signal distortion. Simulation results show that our proposed algorithm can remove most of the binaural colored noise with relatively small signal distortion even under low SNR conditions. Compared to the sophisticated "cocktail-party" processors used for binaural noise reduction, our method is simpler and computationally efficient.