From: Introducing phonetic information to speaker embedding for speaker verification
Ā | Core-extended | 10āsā10ās | ||||
---|---|---|---|---|---|---|
Ā | EER(%) | minDCF08 | minDCF10 | EER(%) | minDCF08 | minDCF10 |
i-vector | 2.02 | 0.0110 | 0.4104 | 10.56 | 0.0530 | 0.9683 |
DNN/i-vector | 1.13 | 0.0055 | 0.2369 | 9.50 | 0.0415 | 0.8697 |
x-vector | 1.45 | 0.0079 | 0.3234 | 9.83 | 0.0431 | 0.9451 |
x-vector-pa (c=0) | 1.34 | 0.0072 | 0.2932 | 8.45 | 0.0459 | 0.8732 |
x-vector-pa (c=0.1) | 1.29 | 0.0061 | 0.2768 | 7.75 | 0.0434 | 0.8627 |
x-vector-mt (3-layer sharing) | 1.18 | 0.0064 | 0.2423 | 8.45 | 0.0399 | 0.8486 |
sc-vector (3-layer sharing) | 1.21 | 0.0062 | 0.2345 | 6.34 | 0.0404 | 0.7746 |
c-vector (c=0.1 + 3-layer sharing) | 0.99 | 0.0049 | 0.2189 | 7.04 | 0.0401 | 0.7782 |