基于参考独立分量分析的语音增强方法

参考独立分量分析（independent component analysis with reference，ICA—R）将源信号的先验知识以参考信号的形式引入学习算法中，可以从混合信号中仅抽取期望的源信号．基于ICA—R提出了一种语音增强新方法．通过比较语音信号和多种噪声信号的特点，合理地构造了具有语音信号重要特性的参考信号，进而应用ICA—R从多种加性噪声中抽取了期望增强的语音信号．计算机仿真和性能分析结果均表明了该方法的有效性．

Speech enhancement method based on ICA-R

Independent component analysis with reference(ICA-R) can extract only some desired source signals from mixtures of all source signals by incorporating prior information into the learning algorithm as reference signals.A novel method for speech enhancement is proposed based on ICA-R.A proper reference signal is first constructed by exploiting different spectrum characteristics between a speech signal and different noise signals,and then used to extract the target speech signal from noisy sounds.Both computer simulation results and performance analyses demonstrate the efficiency of the proposed method.