wuziqi

通过分帧处理用Matlab实现语音增强中的谱减法(Through sub-frame processing using Matlab implementation of spectral subtraction speech enhancement)

很经典的国外模式识别教材MATLAB例程(Classic pattern recognition of foreign materials MATLAB routines)

梯度下降法在一维函数中的优化应用，自己原创，可以运行，无任何错误(Gradient descent algorithm applied in a one-dimensional function optimization, and their own original, can run without error.)

多模板去噪的matlab源代码，对不同模板进行比较(Multi-template denoising matlab source code, a comparison of different templates)

基于受控自回归积分滑动平均 (CARIMA)模型控对象的预测控制程序(Points based on the controlled auto-regressive moving average (CARIMA) Model Predictive Control control procedures)

智能算法研究的书籍，可以研究粒子群的发展以及其现在的研究状况(Intelligent algorithms book, you can study the development of PSO and its current research status)

说明：  根据设计指标提取广义切比雪夫滤波器阶数和传输零点位置，参照《新型微波滤波器的理论与设计》_褚庆昕一书，并提供了滤波器耦合矩阵的计算程序，参照Richard Cameron的方法。(The order of the generalized Chebyshev filter and the position of the transmission zero point are extracted according to the design indicators. Refer to the book "Theory and Design of New Microwave Filters" _ Chu Qingxin, and provide a calculation program for the filter coupling matrix.)

Factorial hidden markov model and compute likelihood per each cycle with train test and validation set. train HMM used Backward-Forward Algorithm.

matlab平面拟合方法，从其他网站上转载来的(matlab plane fitting methods, are reproduced from other sites to the)

说明：  LMS (Least Mean Squares): most basic canonical ANC algo FxLMS (Filtered eXtended Least Mean Squares): adds an additional learned filter for the secondary path signal - signal from cancellation speakers to users ears - to account for phase problems and audio coloration added during practical noise cancellation applications NLMS (Normalized Least Mean Squares): adds simple adaptively updated learning rate on top of LMS to speed convergence. FxNLMS (Normalized Filtered eXtended Least Mean Squares): combines FxLMS and NLMS FuNLMS (Filtered-u Last Mean Squares): adds an additional active LMS filter to FxNLMS to cancel out noise bleeding from the cancellation speaker to the error mic. Fairly robust but convergence not gaurunteed!