Not getting alignment properly even after around 45000 iterations

NVIDIA / tacotron2

Tacotron 2 - PyTorch implementation with faster-than-realtime inference

BSD 3-Clause "New" or "Revised" License

4.97k stars 1.37k forks source link

Not getting alignment properly even after around 45000 iterations #591

Open ov1n opened 1 year ago

ov1n commented 1 year ago

I've been training the model for Sinhala using Sinhala characters. The dataset is in a 22050 sample rate and the symbols have been set properly in the text cleaners.

However, I am not getting proper alignment even after 45k iterations:

Some of my hyperparameters are:

batch_size= 8
learning_rate=1e-4

gujjulassr commented 8 months ago

hi, I'm curious to know if you've made any progress or achieved the results you were looking for?