20080426

基于GMM模型实现语音转换,通过GMM模型对谱包络的建模，实现男女，个性化语音之间转换(Voice conversion using GMM model)

在仿真语音增强时，往往需要向纯净语音中添加适当的噪声然后作仿真对比，这个程序正是实现这了个功能。(When the speech enhancement is emulation, it is often necessary to add appropriate noise to the pure speech and make a simulation contrast. This program is exactly the function of this program.)

VAD去噪算法，可以有效的过滤静音和无效的语音，同时消除了一些背景噪声。(VAD denoising algorithm, can effectively filter invalid mute and voice, while eliminating some background noise.)

语音编码中对LSF系数进行VQ处理，LBG方法生成VQ码书，且采用了新的失真测度(Speech Coding for VQ processing LSF coefficients, LBG method of VQ codebook generation, and adopted a new measure of distortion)

HMM语音识别的matlab程序，举了1-9几个数字的例子，对语音识别的程序进行了验证(HMM Speech Recognition program， cited the example of figures 1-9, the speech recognition process was validated)

基于LMS算法和NLMS算法的自适应声回波消除器AEC，实现的主要功能是：在用户对电视发出的指令混杂了电视扬声器的声学回声的情况下，滤掉电视声音得到较为纯净的用户语音指令信号。(The main function of the adaptive filter based on the LMS algorithm and the NLMS algorithm is mixed: in the instruction issued by the user on the television TV speakers acoustic echo case, the TV sound was removed by filtration to obtain relatively pure user voice command signal.)

绘制一段语音信号的波形并观察其频谱，用三角窗设计一个满足给定相应技术指标的FIR滤波器，对该语音信号进行滤波去噪处理，比较滤波前后的波形和频谱并进行分析。

进行语音合成和分析，最基本的数据来源和可靠合成基础(Speech synthesis and analysis, basic data sources and reliable synthetic base)

基于语音的人类年龄自动估计算法实现，效率比较高(voice-based human age estimation algorithm automatically to achieve more efficient)

attention model for saliency detection.