资源列表
sinusoidal-excitation-linear-prediction-model-meth
- 一种基于正弦激励的线性预测模型的语音转换方法Based on sinusoidal excitation linear prediction model method for voice conversion-Based on sinusoidal excitation linear prediction model method for voice conversion
thesis
- this the document for speech compression-this is the document for speech compression
LRT--VAD
- 利用likelihood ratios模型来进行VAD检测。压缩包中包含了参考文献和所用语音-utilize likelihood ratios test to model the voice activity detection。File includs reference paper and speech which used in the test
bss-sond
- 提出了一种新的自适应盲源分离算法。在无噪音实时两源两传感器的情况下, 一旦观 测信号被白化, 只需要辨识一个特定的旋转矩阵就可以完成盲源分离, 并给出了能表征该旋转矩阵的角的自适应估计器。仿真结果表明, 当满足源峭度和不为零的条件时, 这种方法是一种稳定的和有效的分离算法。-proposes a new adaptive algorithm for blind source separation. In the absence of real-time two noise sources t
hurecogin
- 汉语语音识别中声学建模,非常不错的资料,对了解汉语语音识别技术有帮助-Chinese Speech Recognition acoustic modeling, very good information to the understanding of Chinese speech recognition technology to help
yign
- 语音信号处理，包括时域抽样，进行频谱分析-Including the sampling time domain and frequency spectrum analysis
SpeechSDKDemo
- 微软自带的语言合成和语言识别，语言设别需多次进行声音训练才能提高准确率-Microsoft' s own speech synthesis and speech recognition, language settings do not need to be repeated in order to improve the accuracy of voice training
resmidi
- 用VC编的用于语音的录音和播放，可实时显示。-This software is developed with VC which is mainly used to record and play speech,furthermore,it can be displayed realtime.
voicebox
- 比较经典常用的语音信号处理工具箱函数，可供大家参考-Comparison of the classic common voice signal processing toolbox function, available for your reference
ExtAdvFrontEnd
- vad，语音识别端点检测代码 linux下使用-vad
过零率计算源代码
- 语音端点的检测，不能只以能量的大小进行检测。人声和周边噪音，都可能拥有足够的能量，因此单独能量无法准确检测端点。而过零率在这一点上，则可以起到很大的效果。因此，端点检测时，过零率是不可以缺少的。(Adding zero crossing rate, can accurate detect. Human speech has more zero crossing rate.)
data
- speaker testing data