Design of Paraunitary Oversampled Cosine-Modulated Filter Banks
ICASSP '97 Proceedings of the 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP '97)-Volume 3 - Volume 3
IEEE Transactions on Signal Processing
Signal enhancement using beamforming and nonstationarity withapplications to speech
IEEE Transactions on Signal Processing
Hi-index | 0.00 |
This paper describes a subband adaptive learning algorithm for enhancing microphone array speech signals degraded by a considerable amount of acoustic background noise. The subband multichannel adaptive learning algorithm is adopted to overcome the drawback of slow convergence as well as high computational complexity, which is associated with full band multichannel adaptive LMS algorithm. Simultaneously, oversampled Cosine-modulated filter banks instead of critical sampling filter banks are used to reduce the aliasing effects of subband itself. Simulations experiments show that in addition to fast convergence speed, the proposed microphone array speech enhancement method based on subband adaptive learning algorithm also exhibits a better noise reduction performance than well-known Generalized Sidelobe Canceller (GSC).