search

TaoRuijie / ECAPA-TDNN

Unofficial reimplementation of ECAPA-TDNN for speaker recognition (EER=0.86 for Vox1_O when train only in Vox2)
MIT License
594 stars 113 forks source link

EER 15%? #46

Closed rlaehgud7942 closed 1 year ago

rlaehgud7942 commented 1 year ago

image I got EER:15.98% with pretrain.model and Vox1_O(https://www.robots.ox.ac.uk/~vgg/data/voxceleb/meta/veri_test2.txt) Is it right? image

TaoRuijie commented 1 year ago

It looks wrong...

rlaehgud7942 commented 1 year ago

How long do you think it will take to train this model using the 3070?