摘要
语音增强主要用来提高受噪声污染的语音可懂度和语音质量,它的主要应用与在嘈杂环境中提高移动通信质量有关。传统的语音增强方法有谱减法、维纳滤波、小波系数法等。针对复杂噪声环境下传统语音增强算法增强后的语音质量不佳且存在音乐噪声的问题,提出了一种结合小波包变换和自适应维纳滤波的语音增强算法。分析小波包多分辨率在信号频谱划分中的作用,通过小波包对含噪信号作多尺度分解,对不同尺度的小波包系数进行自适应维纳滤波,使用滤波后的小波包系数重构进而获取增强的语音信号。仿真实验结果表明,与传统增强算法相比,该算法在低信噪比的非平稳噪声环境下不仅可以更有效地提高含噪语音的信噪比,而且能较好地保存语音的谱特征,提高了含噪语音的质量。
Speech enhancement is mainly used to improve the speech intelligibility and speech quality of noise pollution.Its main application is related to improving the quality of mobile communication in noisy environments.Traditional speech enhancement methods include spectral subtraction,wiener filtering and wavelet coefficients.According to the problem of poor speech quality and music noise after speech enhancement for traditional speech enhancement algorithm in complex noise environment,we propose a speech enhancement algorithm combining wavelet packet transform and adaptive wiener filtering.The function of wavelet packet multi-resolution in signal spectrum division is analyzed,and the multi-scale decomposition of noise signal is carried out through wavelet packet.The wavelet packet coefficients of different scales are filtered out by self-adaptive wiener,and then the enhanced speech signal is reconstructed by using wavelet packet coefficients.The experimental simulation shows that this algorithm can not only improve the SNR of speech with noise more effectively under low SNR and non-stationarity noise environment in contrast with the traditional enhancement algorithm,but also can better preserve the spectral features of speech and improve the quality of speech with noise.
作者
董胡
徐雨明
马振中
李列文
任可
DONG Hu;XU Yu-ming;MA Zhen-zhong;LI Lie-wen;REN Ke(School of Information Science and Engineering,Changsha Normal University,Changsha 410100,China;School of Physics and Electronics,Hunan Normal University,Changsha 410181,China)
出处
《计算机技术与发展》
2020年第1期50-53,共4页
Computer Technology and Development
基金
国家自然科学基金(11474090,11174077,61502164)
湖南省教育优秀青年项目(17B025)
湖南省自然科学基金青年项目(2018JJ3557)
长沙师范学院院级教学改革项目(P2016014)
关键词
语音增强
小波包
自适应维纳滤波
多分辨率分析
多尺度分解
speech enhancement
wavelet packet
adaptive wiener filtering
multi-resolution analysis
multi-scale decomposition
