A wonderful ica toollbox about blind source separation for the beginners

G.711语音编解码在CCS上实现的源代码-G.711 voice codec to achieve the CCS source code

用三层小波神经网络实现的与文本无关说话人识别。（识别部分）。输入的是语音特征，输出的是识别结果。训练用的语音特征要事先提取出。-with three wavelet neural network has nothing to do with the text of Speaker Recognition. (Recognition). The admission of voice features, the output is the result of recognition. Training voice features prior to extract.

二相码编码与码组检测，利用matlab实现并仿真-Two-phase coding with code group detection, and to realize the use of matlab simulation

G.729aa 编码器和解码器代码 G.729A 是 G.729 的降低的复杂性标准之一。G.729 是国际电信联盟-T 语音压缩算法，基于共轭结构化代数代码激动的线性预测 (CS-ACELP）。G.729A 压缩 8 kHz 线性音频信号和对他们进行传输在 8 Kbps.输入进行编码为 16 位 PCM。输入码流是分割成 10ms年帧和处理。预测分析做使用合成的筛选器的顺序 10。

语音信号的共振峰抽取的算法-formant extraction in speech signal algorithm

构造一组被噪声污染的方波或三角波数据，试选取适当的FIR滤波器结构和参数，分别应用LMS,DFT/LMS和DCT/LMS算法来实现自适应滤波器，对改组数据进行滤波，并分析仿真结果。-Construction Group was one of the Noise Pollution square or triangular wave data, Examination choose appropriate FIR filter structure and parameters, application LMS, DFT/LMS and DCT/LMS algorithm to achieve adaptive filter, the reorganization data filtering, and analysis of simulation results.

tone encode and decode

改进的BP神经网络的C++源代码,可用于图象识别-Improved BP neural network C++ Source code, can be used for image recognition

AMR-WB声码器VAD（语音活动检测）程序。测试使用8kHz采样率和16位wav量化文件。

对数运算实现是32位的运算包含有指数和对数的求法 -LOG