Speech Sciences (음성과학)

Korean Society of Speech Sciences (한국음성학회)
권호 표지

Modified GMM Training for Inexact Observation and Its Application to Speaker Identification

  • Kim, Jin-Young (Dept. of Electronics and Computer Eng., Chonnam National Univeristy) ;
  • Min, So-Hee (Dept. of Electronics and Computer Eng., Chonnam National Univeristy) ;
  • Na, Seung-You (Dept. of Electronics and Computer Eng., Chonnam National Univeristy) ;
  • Choi, Hong-Sub (Dept. of Electronics Eng., Daejin University) ;
  • Choi, Seung-Ho (Dept. of Multimedia Eng., Dongshin University)
  • Published : 2007.03.31
Download PDF
⟨ Previous Next ⟩

Abstract

All observation has uncertainty due to noise or channel characteristics. This uncertainty should be counted in the modeling of observation. In this paper we propose a modified optimization object function of a GMM training considering inexact observation. The object function is modified by introducing the concept of observation confidence as a weighting factor of probabilities. The optimization of the proposed criterion is solved using a common EM algorithm. To verify the proposed method we apply it to the speaker recognition domain. The experimental results of text-independent speaker identification with VidTimit DB show that the error rate is reduced from 14.8% to 11.7% by the modified GMM training.

Keywords