kdd09transfer.tar

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)

quatization of audio signal and plot the signal after quantization

melpe 编解码 程序 melpe 编解码 程序melpe 编解码 程序 melpe 编解码 程序(melpe decoder)

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.)

说明：  MeLPe Codec 1200 bps

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)

语音信号的矢量量化，其中vector_quantization.m为读取音频文件，转换成语音帧，求lpc及lsf参数以及进行矢量量化的主程序，a_lsf_conversion.m为主程序调用的lpc转lsf参数的子程序，INLS1.wav为音频文件。(Vector quantization of speech signals, wherein vector_quantization.m to read audio file, converted into speech frames, seeking lpc and lsf parameters and vector quantization of the main program, a_lsf_conversion.m main program calls a subroutine parameter lsf lpc turn, INLS1.wav as an audio file.)

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.)

My matlab melp 2400 bps source code for pc

基于麦克风阵列用盖尔圆法进行信源数目估计。输入数据位麦克风采集的信号，阵元数目为10.(Based on microphone array，the number of sources is estimated by Gerschgrid method. Input data is obtained by microphone array and the number of array elements is 10.)