A multichannel subband GSVD based approach for speech enhancement in hearing aids
Ann Spriet, Marc Moonen, Jan M Wouters · 2001
The major goal of a noise reduction algorithm for hear-ing aid applications is to improve speech intelligibil-ity. In [1], a multichannel noise reduction technique, based on a Generalized Singular Value Decomposition (GSVD) has been proposed. The GSVD based filter minimizes the Mean Square Error (MSE) between the desired signal portion in the received signals and a fil-tered sum of the received microphone signals. In this paper, we propose a subband implementation of the GSVD based filter. It is shown that- in case of coloured signals- the subband implementation improves intelli-gibility more than the fullband approach. In addition, a significant complexity reduction is achieved. 1.