论文部分内容阅读
说话人识别属于语音识别的特殊形式,和语音识别一样都是通过对所接收到的语音信号进行处理,提取相应的特征或建立相应的模型,然后再作出判断,说话人识别主要找出包含在语音信号中的说话人的个性因素,强调不同人之间的特征差异。本系统主要使用反射系数、倒谱系数、自相关系数的组合作为特征参数,在此基础上提出了新的矢量量化方法对说话人进行识别。
Speaker recognition belongs to the special form of speech recognition. Like speech recognition, speech recognition is processed through the received speech signal, the corresponding feature is extracted or the corresponding model is established, and then the judgment is made. The speaker recognition mainly finds out the content contained in The speaker’s personality in speech signals emphasizes the differences in characteristics among different people. The system mainly uses the combination of reflection coefficient, cepstral coefficient and autocorrelation coefficient as the characteristic parameters. Based on this, a new vector quantization method is proposed to identify the speaker.