集成电路设计与集成系统毕业论文.docVIP

题目：自适应语音去噪系统研究 学 院 信息科学与工程 专 业 集成电路设计与集成系统  摘 要 随着信息处理理论和技术的发展，滤波技术的应用日益广泛。传统数字滤波器由于权系数固定，在实际应用中缺乏足够信息去计算固定系数，难以解决很多实际问题。自适应滤波技术的出现克服了这一问题，其显著特点是无需用户干预即能自动改变响应以改善性能。因此，自适应滤波算法的研究已成为当前自适应信号处理中最为活跃的研究课题之一。自适应滤波技术包括：自适应信道均衡、信号增强、自适应噪声抵消等。本文以研究自适应噪声抵消技术为目的，利用MatLab仿真Least mean square（LMS算法）和小波阈值算法的去噪效果并进行比较。纵向方面，在不同的噪声环境下，分别使用LMS和小波阈值算法进行去噪处理，分析两种算法在不同的噪声环境下的去噪效果；横向方面，在同一噪声环境下，分别使用LMS和小波阈值两种算法进行去噪处理，比较两种算法在同一噪声环境下去噪性能的优劣性。结果表明，本文研究的LMS去噪算法的去噪效果较好，能有效抑制各种噪声污染，提高信噪比60%～75%。此外，LMS算法更简洁，对于噪声的适应性也比小波去噪更强。 关键词：自适应滤波技术；自适应噪声抵消；LMS算法；小波去噪； Abstract With the development of the information processing theory and technology, filter technique is applied more and more extensively. The traditional digital filter, owing to a fixed weight coefficient which in practice lacks sufficient information to calculate, is unable to solve as many practical problems as possible. Adaptive filtering technology changed this situation. Without the intervention of users, it can automatically adapt itself to improve its performance. Adaptive filtering algorithm, hence, becomes one of the most attractive researches in present adaptive signal processing field. Adaptive filtering technology includes adaptive channel equalization, signal enhancement, adaptive noise cancellation, etc. This paper aims to do research on the adaptive noise cancellation system through testing the denoising effects, processed respectively by Least mean square algorithm(LMS algorithm) and the wavelet threshold denoising algorithm, in MatLab and then comparing the effects. Vertically, in different noise environments, denoise with the two algorithms and analyse their denoising effects. Horizontally, in the same noise environment, denoise separately with the two algorithms and compare their strengths and weaknesses. It turns out that the LMS algorithm，which this paper targets on，has a better quality in terms of denoising effectiveness, suppression of noise pollution