Speaker identification based on EMD
Yali Liu, Hongwu Yang, Hui Zhou · 2009
This paper proposes a novel approach which combines empirical mode decomposition (EMD), short-time analysis and support vector machine (SVM) for text-independent speaker recognition. Short-time analysis is used for the result of empirical mode decomposition to extract speech features of speakers, and then the support vector machine are used for speaker recognition. Experiments demonstrate that the proposed approach outperforms GMM based traditional methods, with the increased recognition rate from 92.5% to 95.1%.