From: Text-independent speaker recognition based on adaptive course learning loss and deep residual network
Dataset
# of speakers
# of utterances
# of pair
VoxCeleb2 dev
5994
1,092,009
-
VoxCeleb1 dev
1211
148,642
VoxCeleb1 test
40
4715
37720
VoxCeleb1-E test
1251
145,375
581,480
VoxCeleb1-H test
1190
138,137
552,536