摘要
将奇异谱分析理论引入小波阈值算法中实现了一种基于奇异谱分析的自适应最优分解层数确定算法。通过对比不同信噪比下带噪信号的奇异谱分布情况,根据小波系数的奇异谱特性来确定最优分解层数。经测试,该算法可以根据带噪信号受噪声干扰情况自适应地确定最优分解层数,有效提高了语音增强效果并且避免了不必要的硬件资源浪费。
This paper, applying the singular spectral analysis theory to the wavelet threshold denoising algorithm, realizes an adaptive determination algorithm for optimal level of decomposition. Comparing distribution of the singular spectral with noise signals under different SNR environments, it determines the optimal decomposition level based on the singular spectrum characteristics of wavelet coefficients. Being tested, the algorithm can adaptively determine the optimal decomposition level, according to noise interference of noise signals, effectively increasing voice enhancement effect and avoiding unnecessary waste of hardware resources.
出处
《计算机工程与应用》
CSCD
2012年第36期137-141,共5页
Computer Engineering and Applications
基金
淮南市科技局项目(No.2012A01003)
淮南师范学院科研项目(No.2012LK10)
关键词
最优分解层数
语音增强
奇异谱分析
阈值降噪
optimal decomposition level
speech enhancement
singular spectral analysis
threshold denoising
