PyTorch implementation of "Generalized End-to-End Loss for Speaker Verification" by Wan, Li et al.
BSD 3-Clause "New" or "Revised" License
575
stars
166
forks
source link
torch.norm will give a scalar, Is this a bug? #16
Closed
fanlu closed 5 years ago
https://github.com/HarryVolek/PyTorch_Speaker_Verification/blob/29232c8e3512b576b8c60b7b5aacea578c1e1369/speech_embedder_net.py#L32 torch.norm will give a scalar, and In paper eq(4), I think x should be normalized by each embedding. F.normalize(x, dim=-1, eps=1e-6), Am I right?