摘要
抽取最佳鉴别特征是说话人辨认中的重要一步。本文在使用美尔倒谱系数(MFCC)及一阶差分组成的特征参数的基础上利用主分量分析(PCA)和线性判决分析(LDA)结合的提取方法,构造了一种新的特征参数。这种新的参数具有最佳鉴别特性,然后用支持向量机(SVM)对提取的特征分类辨认。实验结果表明该方法能更好地识别说话人,有更好的识别能力。
Extracting the best distinction characteristic is one important step in the speaker identification. This article uses the method which combines the primary component analysis (PCA) and the linear discriminate analysis (LDA) to transform the characteristic based on Mel frequency cepstral coefficients (MFCC), has constructed one kind of new characteristic parameter. This kind of new parameter has the best discriminate characteristic, and then it uses support vector machines (SVM) to distinguish the extracted characteristic. The experimental result indicates that this method can distinguish speaker better,and has the better identification ability.
出处
《电子测量技术》
2008年第9期126-128,共3页
Electronic Measurement Technology