yuyinshiyupinyufenxi

基于DTW算法实现孤立数字及连续数字语句的识别，识别率还可以（内含语音数据集）(Based on DTW algorithm, isolated number and continuous number statements can be recognized, and the recognition rate can also be (including voice data set))

编写一Matlab函数，用30ms三角窗和15ms帧移计算语音信号的STFT。并用OLA法重建原始信号。设计一种基于OLA的综合方法，以通过重复每帧对语音信号以因子2进行时域扩展。(Write a Matlab function, STFT and 30ms triangular window and 15ms frame-shift calculation of the speech signal. And the OLA method to reconstruct the original signal. Design an integrated approach based on OLA, by repeating each frame of the speech signal by a factor of 2 when the domain extension.)

基于谱减法的语音降噪处理，里面有声音文件，可以直接运行观察效果(Based on the spectral subtraction speech denoising processing, there are sound files, can be run directly observed effect)

基于LMS算法的变换域的源程序，里面给出了学习曲线，迭代后的输出曲线(LMS algorithm based on the Domain source, which is a learning curve, the output iterative curve)

voice recognition using mfcc

比较经典的用lms算法实现自适应均衡器的matlab的程序，适合初学者学习(Comparison of the classical LMS algorithm with adaptive equalizer matlab procedures, suitable for beginners to learn)

这是一个很好的频域分块自适应滤波的程序，应用于回声消除上，并与NLMS自适应滤波做了运行速度的比较,FLMS比NLMS快几十倍。(This is a very good frequency-domain block adaptive filtering procedure applied to echo cancellation, and comparison with the NLMS adaptive filter for speed, FLMS several times faster than the NLMS.)

完整的基于HMM的语音识别系统，时间序列数据分析中的梅林变换工具，最终的权值矩阵就是滤波器的系数。( Complete HMM-based speech recognition system, Time series data analysis Mellin transform tool, The final weight matrix is ??the filter coefficient.)

本程序音频共有四个，借助鸡尾酒会问题思想，用ica算法实现四个音频的分离(The audio program for a total of four, with the help of the cocktail party problem, realizing the separation of four audio using ICA algorithm)

基于LMS(最小均方误差算法)的自适应滤波 基于LMS(最小均方误差算法)的自适应滤波(based on the LMS (MMSE) algorithm based on the LMS adaptive filtering (minimum mean square error algorithm) the adaptive filtering based on the LMS (minimum mean square error algorithm) Adaptive Filter)