采用期望最大算法最优化初始高斯混合模型的程序，笔者借用ubm自适应方法，很好的解决了模型不收敛的问题。-expectations largest algorithm using optimization initial Gaussian mixture model procedures, the author borrowed ubm adaptive method, a good model is not the solution convergence problem.

快速块LMS算法 利用FFT，实现频域的LMS快速算法，并且能够加快算法的收敛速度-Fast Block Least Mean Squares

块LMS算法 相对于逐点计算的LMS算法，块LMS算法具有相对较高的计算效率，同时在一定程度上提高了算法的收敛稳定性-LMS algorithm block-by-point LMS algorithm for computing the block LMS algorithm has a relatively high computational efficiency, while a certain extent, to improve the stability of the algori