A New Speech Enhancement System Based on the Adaptive Thresholding of Wavelet Packets

In this paper, we propose a new speech enhancement system using wavelet thresholding algorithm. In the proposed system, we introduce a new algorithm which does not require the voiced/unvoiced detection process. Also an adaptive threshold value and a new thresholding function are introduced to improve the speech enhancement performances and speech auditory quality. A voice activity detector (VAD) algorithm is designed for this system to update noise statistics. The proposed method is evaluated on several speakers and various noise conditions including white Gaussian noise (WGN), pink noise, and multi-talker babble noise and the performance results are reported.