49040380LPC

基于melp算法的语言编码传输，用于语言压缩传输(Speech coding transmission based on MELP algorithm)

说明：  线性预测编码（LPC）是主要用于音频信号处理与语音处理中根据线性预测模型的信息用压缩形式表示数字语音信号谱包络（en:spectral envelope）的工具。它是最有效的语音分析技术之一，也是低位速下编码方法高质量语音最有用的方法之一，它能够提供非常精确的语音参数预测。(linear predictive coding)

假设一段语音信号，用其作为序列，编程实现LBG算法产生码书。(Assuming a speech signal, use it as a sequence, programming the LBG algorithm to generate the codebook.)

这是G.723和G.729的音频编解码的源代码，当在网络带宽不够的时候可考虑采用该语音编码方式。(This is the G.723 and G.729 audio codec source code, When the network bandwidth may be insufficient time to consider the use of the voice coding.)

TXT信号的获取：一般都是拿到一个.WAV的信号，用cooledit或者其他的打开，另存为txt格式，注意的是，此时txt文件的前面两行会写入一些采样率的信息，将这几行删掉，程序就可以顺利得读取你保存的txt了。(TXT signal acquisition: generally get a .WAV signal, with coolit or other open, save as txt format, note that at this time txt file in front of two lines will write some sampling rate information, will These lines deleted, the program can be smooth to read your saved txt.)

MMSE语音降噪算法，使用维纳滤波器的方法，运算量适中，能实时处理(MMSE weiner noise reduction)

A率13折线的编译码实现 1、了解PCM及13折线A率编码及译码的原理； 2、随机给出一个语音信号，并用A率13折线PCM对其编码； 3、将编码后的信号经过加性高斯白噪声信道，并在接收端对其进行译码； 4、分别画出原始语音信号、PCM编码信号以及译码后信号的波形。 (A rate of 13 polyline encoding and decoding implementation 1, to understand the principles of PCM and the broken line 13 A rate of encoding and decoding Two were given a voice signal, and PCM encode A rate of 13 polylines 3, the encoded signal through an additive white Gaussian noise channel, and the receiver decodes 4, respectively, to draw the waveform of the original speech signal, the PCM encoded signal and decoding signals.)

说明：  光谱预处理auto 光谱预处理center(Spectral pretreatment)

迁移学习程序，适用于NLP问题，大家可以看看。(To migrate learning program suitable for NLP problems)

双通道语音增强算法，消除环境噪声。采用归一化自适应方法，噪噪声抵消10dB，语音保持较好可懂度。(The dual-channel speech enhancement algorithm to eliminate ambient noise. Using normalized adaptive noise noise offset by 10dB, and the voice to maintain good intelligibility.)