最终代码说话人识别

code for voice analysis final code for sapeech recognition

利用matlab提取声音的基本特征，包络、MFCC、声谱图、能量。代码不复杂，但是省去大家找这些的时间。(The basic feature extraction using matlab sound envelope, MFCC, sonograms, energy. Code is not complicated, but we look for these omitted time.)

语音识别的hmm技术，对做语音识别的朋友很有帮助 (Hmm speech recognition technology, speech recognition make friends very helpful)

语音信号处理中用于说话认识别中预处理部分的分帧函数代码(Speech signal processing for speech understanding in other parts of the sub-frame in the pre-function code)

时间延迟估算是在现供求信号处理中经常会被用到,以检测波入射的方向(雷达、声纳和地震波)。这里给出一个自适应时间延迟估算--Adaptive(LMS) Time Delay Estimation--的MATLAB程序，供大家参考。(Time delay estimate is that in the current supply and demand signal processing are often used to detect the direction of the incident wave (radar, sonar, and seismic waves). Here are a self-adaptive time delay estimation Adaptive (LMS) Time Delay Estimation of MATLAB procedures for your reference.)

声纹识别系统，顾名思义就是能够通过说话人的声音特点来判断说话人的一个系统。(Voiceprint recognition system, as the name suggests is to a system by the sound characteristics of speaker to judge the speaker.)

基于倒谱短时部分反映了语音的声道特性，先用汉明窗取一帧语音，然后经变换得到语音倒谱，将倒谱短时部分取出，进行正交反变换后将得到声道的对数谱，即得到语音频谱的包络。将频谱包络和频谱画在一张图上，有很好的对比效果。获取的包络效果十分好。(Based Cepstral partly reflects the short channel characteristics of the speech, first take a Hamming window with a frame of speech, and speech cepstrum obtained by converting the cepstrum short segment out inverse orthogonal transform to obtain channel will of the spectrum, i.e. to obtain the envelope of the speech spectrum. The spectral envelope and spectral painted on a chart, there is a good contrast. Get the envelope effect is very good.)

很好的一本关于语音信号处理的书，而且是中文版的，比较好用(A good book on speech signal processing, and is the Chinese version, more useful)

PESQ-主观语音质量评估。 ITU-T P.862建议书提供的客观MOS值评价方法。PESQ可以广泛的用于现场和模拟网络中端到端测量。背景(环境)噪声和噪声处理,可以通过用PESQ比较干净的、未处理的信号和编码的、经噪声干扰失真的信号评估得到。(PESQ- subjective speech quality assessment. ITU-T P.862 recommendations for objective MOS value uation method book provides. PESQ can be widely used in the field and in the end measurement analog networks. Background (ambient) noise and noise processing, you can use PESQ relatively clean, unprocessed signal uation and signal encoded by the noise distortion obtained.)

采用加权网络中节点强度和权重都是幂率分布的模型，计算两个矩阵之间的欧氏距离，实现了对10个数字音的识别。( Using weighted model nodes in the network strength and weight are power law distribution, Calculation of the Euclidean distance between the two matrices, To achieve the recognition of 10 digital sound.)