摘要
介绍了一种基于峭度的盲源分离算法,利用峭度极大来度量极大化非高斯性,通过渐进正交化的不动点迭代找到独立成分,并对带噪多人声混叠语音信号进行分离仿真,从而提取出感兴趣的目标语音,验证了该算法的可行性;通过与其他盲源分离算法的分离结果进行定性和定量的对比分析,验证该算法的有效性和应用前景。
Introduce a method based on kurtosis blind source separation algorithm, using kurtosis maximization to measure Maximization of Nongaussianity, then Through Gram-Schmidt orthogonalization fixed-point iteration to find the independent components,and with noisy model of the separation of voice signal simulation, thus extracted interested target voice to verify the feasibility of the algorithm; By comparison with other BSS algorithm of separation result are the quantitative and qualitative analysis, verify the effectiveness of the algorithm and application prospect.
出处
《信息通信》
2011年第1期29-32,共4页
Information & Communications
基金
河南工程学院青年基金资助项目
关键词
峭度
盲源分离
极大化非高斯性
Kurtosis
BSS（Blind Source Separation）
Maximization of Nongaussianity
