search

HarryVolek / PyTorch_Speaker_Verification

PyTorch implementation of "Generalized End-to-End Loss for Speaker Verification" by Wan, Li et al.
BSD 3-Clause "New" or "Revised" License
578 stars 164 forks source link

Question on time cost during each iteration by uis-rnn #19

Closed hcfeng201 closed 5 years ago

hcfeng201 commented 5 years ago

Sir, I've trained the d-vectors by your code, but it takes much more time than UIS-RNN's own test sequence which have the same sequence length and dims during each iteration. The phenomenon is just like what I described in the following question. Thank you in advance. https://github.com/google/uis-rnn/issues/27

HarryVolek commented 5 years ago

UIS-RNN performs a different task than this project. They recommend the following. Here test_sequence should be a 2-dim numpy array of type float, corresponding to a single observation sequence.