ncm

This is program set for SLAM using Matlab and filtering method. This code is fast. It is pretty good for student and teacher to learn. It is also easy to improve based on this code. It is really helpful for the new guy in this area. This is a platform and easy to extend using other filtering methods

快速傅里叶变换， 频谱的泄露，混叠误差实例的matla程序。(Fast Fourier Transform: In order to reduce the number of DFT calculations of a fast algorithm.)

主要是些matlab的常用指令，对大家可能会有所帮助(Mainly those commonly used matlab instructions may help to everyone)

通过matlab语言编程，实现利用Gauss消去法求解线性方程组(Through matlab language programming, using the Gauss elimination method for solving linear equations)

任务： 1） 借助MATLAB画出误差性能曲面和误差性能曲面的等值曲线（参考PPT2.1第17页的两幅图）； 2） 写出最陡下降法以及LMS算法的计算公式（取 ）； 3） 用MATLAB产生方差为0.05, 均值为0白噪音S(n)，并画出某次采样得到的波形（即产生任意一个噪声随机序列）； 4） 根据 2）中的公式，并利用 3）中产生的S(n)，在 1）中的误差性能曲面的等值曲线上叠加画出采用最陡下降法以及LMS法时H(n)的在叠代过程中的轨迹曲线（参考PPT2.1第17页的右下图的曲线1和曲线2）。 5）用MATLAB计算并画出LMS法时 随时间n的变化曲线（对应S(n)的某一次的一次实现）和e(n)波形； 注意：某一次实现的结果并不能从统计的角度反映实验的结果 的正确性，为得到具有统计特性的实验结果，可用足够多次的 实验结果的平均值作为实验的结果。用MATLAB计算并画出 采用LMS法时，J(n)的100次实验结果的平均值随时间n的变 化曲线（即 生成随机噪声信号并计算结果，重复执行100次， 求平均结果）。 6）在 1）中的误差性能曲面的等值曲线上，叠加画出采用LMS法得到的100次实验中的H(n)的平均值的轨迹曲线； (Task: 1 ) Draw with MATLAB error performance surface and surface contour error performance curve ( refer PPT2.1 two chart on page 17 ) 2 ) Write the steepest descent method and the LMS algorithm formula ( take ) 3 ) Using MATLAB to generate variance of 0.05 with a mean of 0 white noise S (n), and draw a particular sampling waveform ( ie, any noise generated random sequence ) 4 ) based on 2 ) of the formula , and use 3 ) generated in S (n), in a ) the error performance curves superimposed on the surface contour plot using the steepest descent method and the LMS method when H (n) the iterative process in the trajectory curve ( refer PPT2.1 page 17 bottom right of the curves 1 and 2 ) . 5 ) using MATLAB LMS method is used calculate and draw the curve with time n (corresponding to S (n) of a first implementation of a ) and e (n) waveform Note : in a first implementation of the results are not reflected from a statistical point of view of accuracy of the experimental results ,)

使用程序实现mse函数的三维设计实现，可以表现出不同的最小均方误差的各种情况(MSE realize the use of procedures designed to achieve three-dimensional function, can be expressed in different MMSE various situations)

PDF documents about speaker verfication

神经网络中用矩阵分块的方法处理数据量过大的情况，效果很好(Neural network is used in the method of matrix blocking deal with the situation of the large amount of data, the effect is very good )

this code extract the feature from speech signal in many algorithms

说明：  介绍使用matlab对ofdm进行编程，掌握其基本原理(Introduction to program using matlab on ofdm and grasp the basic principles)