The Filtering Technique With Wavelet Coefficient Threshold For Hearing Aids System Background Noise Reduction

• Main Authors: Guo-Zheng Wong, 翁國正
• Other Authors: Cheng-Chi Tai
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/85154384716356669111

碩士 === 國立成功大學 === 電機工程學系碩博士班 === 91 === In this paper, a noise reduction technique that implemented with the TMS32C6711DSK system for digital hearing aids is described. The technique uses an algorithm based on a wavelet coefficient threshold method to reduce the background noise. To eliminate the noise in the speech signals, an analyzable threshold is given from the Wavelet toolbox of MATLABTM. In the process of noise reduction, the setting of an optimal threshold relies on statistical data from experiences. If real-time process is required in the hardware (TMS32C6711DSK) implementation, the de-noising algorithm based on a choice of the Haar’s wavelets is recommended for that it has simpler computing. The results show that the system can reserve the most of human speech signal in sub-band decomposition process. When the noise and speech signal are amplified at the same time, hearing aid users can’t clearly hear speech signal in noisy environment. By using the wavelet coefficient threshold, we can discriminate the difference between speech signal and noise.