该文章是语音去噪方面的，比较新，希望对大家有帮助

符合G。729A标准的语音压缩源代码， 很经典，未经过优化含有错误，请修改后使用，-with G. 729A standard voice compression source code, classic, not optimized to contain errors, please use the revised,

说话人识别程序,VQ识别方法,特征参数是MF-Speaker recognition procedures, VQ identification methods, feature parameters are MF

Sound separation by ICA

一个简单的报时小软。可以中文报时 初学语音识别，希望同样爱好的朋友可以多多交流-A simple small-time soft. Time learning in Chinese speech recognition, hope that the same lot of loving friends can exchange

输入WAV文件并获取特征（*.txt）、单声道或立体声，然后按照以下步骤从WAV文件中提取特征，

语音识别与合成入门2，word或压缩包中内有matlab源码-speech recognition and synthesis 2 entry, word or compressed packages within Matlab FOSS

这是一个在语音识别中求Mel倒谱系数的matlab程序。 这个程序里面还需要用到几个其他的函数，都在VoiceBox工具箱里面，可以到网上下载-Calculate the mel cepstrum of a signal，The program also needs to be used in several other functions, in VoiceBox toolbox, can to download

自适应线性预测器，采用FIR滤波器实现。-adaptive linear predictor, the use of FIR filters achieve.

本软件是一套完全基于神经网络的语音识别处理软件。 本软件为纯绿色软件，无须安装，解压到任意一个目录下，直接运行主程序 whd.exe ，卸载时删除所在目录即可，软件不会给系统留下任何垃圾，请放心试用，但为了您的数据安全，在使用时请先将重要数据保存备份，呵呵，我信不过微软啊:) 本软件使用的是本人自行开发的神经网络处理系统，但因工程浩大，且本人穷，只能独立编写，所以现在这个试用版所使用的神经网络系统并不完整，只实现了我最初设计的一半功能，识别效果还不是很理想，正在改进中:) 但还是希望大家能对软件提出宝贵意见，我的邮箱：whd_china@etang.com 或 yinshishan@sina.com -this a completely software based on neural network speech recognition processing software. The software for the pure green software, no installation, to extract an arbitrary directory and run the main program whd.exe. unloading directory will be deleted, the software system will not leave any garbage, please rest assured that trial. But for the sake of your data security, the post will be used to preserve important data backup, Oh, I trust Microsoft ah : ) This is the software I use to the development of the neural network processing system, but the project is a huge one, himself poor, only independent prepared, the trial version now used by the neural network system is incompl

Speech Quantization_Compare SNRs of speech cs5 with mu-law companding and uniform quantization -Speech Quantization_Compare remnants of spee ch cs5 with mu-law companding and uniform quant ization