Abstract
Although numerous speech representations have been reported to be useful in speaker recognition, there is much less agreement on which speech representation provides a perfect representation of speaker-specific information. In this paper, we charaterize a speaker's identity through the simultaneous use of various speech representations of his/her voice. We present a parametric statistical model, generalized Gaussian mixture model, and develop an EM algorithm for parameter estimation. Our approach has been applied to speaker recognition and comparative results on KING corpus demonstrate its effectiveness. © Springer-Verlag Berlin Heidelberg 2004.
| Original language | English |
|---|---|
| Title of host publication | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)|Lect. Notes Comput. Sci. |
| Publisher | Springer Nature |
| Pages | 592-599 |
| Number of pages | 7 |
| Volume | 3072 |
| Publication status | Published - 2004 |
| Event | International Conference on Biometric Authentication - Hong Kong Duration: 1 Jan 1824 → … |
Conference
| Conference | International Conference on Biometric Authentication |
|---|---|
| City | Hong Kong |
| Period | 1/01/24 → … |