ide8 / tacotron2

Multispeaker & Emotional TTS based on Tacotron 2 and Waveglow

ide8/tacotron2 Issues

Loss is NaN on first step, if data exceeds 100 sentences
Updated a year ago3
Thank you for this repo. I trained for 300 epoch. I am getting the audio output but its noisy. Can you suggest me how to get better audio output from this model?
Updated a year ago
Dataset
Closed 3 years ago1
Pretrained Model
Closed 3 years ago
any related papers?
Closed 3 years ago
Inference inconsistency
Closed 3 years ago1
Griffin Lim
Closed 4 years ago1