基于特定听觉特征的盲信号提取方法
Blind signal extraction based on specific auditory feature

针对参考信号波形难以构造等情况，提出一种利用期望源信号听觉特征进行信号盲提取的方法，根据源信号的听觉特征先验知识，将提取信号与期望信号听觉特征向量的接近程度作为度量，使用非线性最小二乘法进行迭代计算分离向量，提取出与期望信号在听觉特征上最为接近的源信号。对这种方法进行了理论推导，并使用不同类型的信号进行了仿真。仿真结果表明，该方法具有良好的效果，且由于仅利用了信号的听觉特征，而不需要参考信号的具体波形，对于不同类型的信号，具有广泛的适应性，具有良好的应用价值。
In the case that the reference signal waveform is difficult to provide, a blind extraction method based on auditory feature is proposed in this paper. According to the known auditory feature, the accessibility of auditory features between expected signal and extracted signal is treated as a measure, the nonlinear least square algorithm is used to calculate the separation vector in order to extract the source signal that is closest to the expected signal in auditory feature. The theory of the method is described in this paper, and the simulation is also performed with different kinds of signals. The simulation result shows the effectiveness of the method, which needs no waveform but only the auditory feature of reference signal. And, the method is adaptive to different kinds of signals and has a good performance in practical applications.