小波在人工耳蜗语音处理中的应用。

Application of wavelet in speech processing of cochlear implant.

作者信息

Tian Guan, Shuli Yu, Datian Ye

机构信息

Dept. of Biomedical Engineering, Tsinghua Univ., Beijing, 100084. Email:

出版信息

Conf Proc IEEE Eng Med Biol Soc. 2005;2005:5339-42. doi: 10.1109/IEMBS.2005.1615687.

DOI:10.1109/IEMBS.2005.1615687

PMID:17281457

Abstract

Cochlear Implant bypasses damaged hair cells and stimulates the auditory nerve directly in the cochlea. Wavelets are especially useful for speech processing of CI, since in the case of wavelets one speaks of a constant-Q filter bank as well as the cochlea. In this paper, a continuous wavelet has application in the filter bank design and an effective result is obtained. Also the lifting scheme, the so-called second generation of wavelet, is adopted, combined with pitch extraction algorithm, to extract fundamental frequency of speech signal. Veracious result is obtained, which shows the great potential of wavelet an lifting scheme in the application in speech processing strategy for cochlear implants based on characteristics of Chinese language.

摘要

人工耳蜗绕过受损的毛细胞，直接在耳蜗中刺激听觉神经。小波对于人工耳蜗的语音处理特别有用，因为在小波的情况下，可以将其视为一个恒定Q滤波器组以及耳蜗。在本文中，连续小波在滤波器组设计中得到应用并取得了有效成果。此外，还采用了提升方案，即所谓的第二代小波，并结合基音提取算法来提取语音信号的基频。获得了准确的结果，这表明基于汉语特征的小波提升方案在人工耳蜗语音处理策略应用中具有巨大潜力。