search

auspicious3000 / autovc

AutoVC: Zero-Shot Voice Style Transfer with Only Autoencoder Loss
https://arxiv.org/abs/1905.05879
MIT License
983 stars 207 forks source link

Tranning is too slow #23

Closed 1015720437 closed 4 years ago

1015720437 commented 4 years ago

I used a training set of 400 speaker each of them have 350 senten .Training is very slow. is that too big? What is the training set you use? Can you tell me?

auspicious3000 commented 4 years ago

VCTK. You can find all these details in the paper.

1015720437 commented 4 years ago

VCTK. You can find all these details in the paper.

How many epochs the model ran in the demo. My training is very slow, it accounts for 17G gpu memory. But in the case of reducing the data set, one still takes an hour. I saw that you wrote epochs in the hyperparameter file is 3000. How long did it take you to train?

auspicious3000 commented 4 years ago

Training should be very fast using pytorch and takes very little memory.