search

jaywalnut310 / vits

VITS: Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech
https://jaywalnut310.github.io/vits-demo/index.html
MIT License
6.91k stars 1.27k forks source link

Question about VITS KL Loss Formula #112

Open MMingabc opened 1 year ago

MMingabc commented 1 year ago

Thank you for this fantasic work. I think the kl loss in VITS might have missed one term. See: https://statproofbook.github.io/P/norm-kl.html#mjx-eqn-eq%3Anorm-KL

lexkoro commented 1 year ago

https://github.com/jaywalnut310/vits/issues/6#issuecomment-861903556