LPC倒谱参数的说话人特征分析

通过对多个说话人与文本无关的发音材料进行统计分析，比较了ＬＰＣ倒谱的不同分量对说话人特征区分所作的贡献，研究了ＬＰＣ倒谱所体现的说话人共性及个性特性，同时使用了Ｆｉｓｈｅｒ判据，最近邻错误率，超球错误率、类内，类间平均距离等量度方法，从不同角度探讨了ＬＰＣ倒谱参数的说话人统计特征，并考查了它在正交变换，最大可分变换下的变化及以不同距离作为测量手段所产生的不同效果，这些工作是对ＬＰＣ倒谱参数用于说话

LPC-Cepstrum Analysis of Speaker Features

LPC-cepstrum is successfully used in many speaker recognizers. This paper investigates different components of the LPC-cepstrum parameters devoted to discriminative features of speakers and gives a describtion of general and individual characteristics of speakers based on the statistical analysis of a variety of text-independent speech databases from several speakers. The statistical features of speaker's LPC-cepsturm are studied from various points of view,such as fisher-ratio,nearest-neighbour error,superball error and ratio of the between-class and within-class scatter measure and its change is examired under orthogonal transform or max-discriminative transform. The conclusions result in further understanding of an internal cause of LPC-cepstrum used in speaker recognition and will be of great value to the effective selection of speaker discriminative features,a significant reduction of feature dimensionality and the improvement of the recognizer performance.