论文部分内容阅读
Ｗｉｎｄｏｗｓ的ＷＡＶ文件提供的语音信息可以满足语音特征识别的要求．根据讲话者在讲话时鼻腔相对固定以及汉语鼻音声母少的特点，使用极零模型，并以所有汉语鼻声母音节的零点谱为参数，实现汉语讲话音辨别．所选择的参数有效，系统模板存储量少，系统训练和识别的运算量小．
Voice information provided by WAV files in Windows can meet the requirements of voice feature recognition. According to the fact that the nasal cavity is relatively fixed and the nasal consonant is less in the speaker’s speech, the zero-zero model is used and the zero-point spectrum of all Chinese nasal consonants is used as the parameter to distinguish Chinese speech. The selected parameters are valid, the system template storage is small, the system training and identification of a small amount of computation.