Abstract | ||
---|---|---|
We propose a strategy for discriminative training of the i-vector extractor in speaker recognition. The original i-vector extractor training was based on the maximum-likelihood generative modeling, where the EM algorithm was used. In our approach, the i-vector extractor parameters are numerically optimized to minimize the discriminative cross-entropy error function. Two versions of the i-vector extraction are studied the original approach as defined for Joint Factor Analysis, and the simplified version, where orthogonalization of the i-vector extractor matrix is performed. |
Year | Venue | Keywords |
---|---|---|
2011 | 12TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION 2011 (INTERSPEECH 2011), VOLS 1-5 | speaker verification, i-vectors, PLDA, discriminative training |
Field | DocType | Citations |
Speaker verification,Error function,Pattern recognition,Computer science,Matrix (mathematics),Expectation–maximization algorithm,Speech recognition,Speaker recognition,Artificial intelligence,Extractor,Orthogonalization,Discriminative model | Conference | 3 |
PageRank | References | Authors |
0.45 | 10 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ondřej Glembek | 1 | 852 | 64.75 |
Lukas Burget | 2 | 169 | 14.43 |
Niko Brümmer | 3 | 595 | 44.01 |
Oldrich Plchot | 4 | 327 | 30.88 |
Petr Schwarz | 5 | 991 | 69.47 |