Rayhane-mamah / Tacotron-2

DeepMind's Tacotron-2 Tensorflow implementation
MIT License
2.28k stars 905 forks source link

Try to understand alignment plot #390

Open zhang-jian opened 5 years ago

zhang-jian commented 5 years ago

Hi,

I noticed that the alignments from a trained model has some fuzzy parts as circled in the image. step-22500-align

I couldn't understand (1) what caused this? (2) How to interpret these parts, are these parts badly aligned or it is normal? Just wondering if anyone studied this before and can give me some directions, thanks.