中文说明：基本的声强向量估计法无法适用于多声源情况下的方位估计，但由于语音信号具有时频稀疏性，利用该特性则可实现多语音源时的方位估计，其中典型的方法如直方图法。该方法首先计算各声源瞬时方位的直方图分布，然后通过搜索直方图的峰值位置得到目标声源的方位估计。直方图方法在低噪声混响条件下具有良好的性能，但随着环境噪声和混响的增大，其性能恶化。此外，由于没有充分利用语音信号的稀疏性条件，其对阵列尺寸的选择也较为敏感
English Description：
The basic sound intensity vector estimation method can not be applied to the DOA estimation in the case of multiple sound sources. However, due to the time-frequency sparsity of speech signal, the DOA estimation in the case of multiple sound sources can be realized by using this characteristic. The typical method is histogram method. Firstly, the histogram distribution of instantaneous azimuth of each sound source is calculated, and then the azimuth estimation of the target sound source is obtained by searching the peak position of the histogram. Histogram method has good performance in low noise reverberation, but its performance deteriorates with the increase of ambient noise and reverberation. In addition, the sparsity condition of speech signal is not fully utilized, so it is sensitive to the selection of array size