[1] Chung, J. S., Nation, D., & Nagrani, A. (2018). VoxCeleb2: Deep Speaker Recognition. In Proc. Interspeech 2018 (pp. 272–276).

The vox-cpk.pth.tar file likely contains pre-trained model weights for speaker verification using the VoxCeleb2 dataset. These weights can be used for tasks such as speaker identification, verification, or clustering.

References: