NVIDIA / tacotron2

Tacotron 2 - PyTorch implementation with faster-than-realtime inference
BSD 3-Clause "New" or "Revised" License
5.11k stars 1.39k forks source link

How to set Tacotron2 for Bengali language? #492

Open nafeesmahbub opened 3 years ago

nafeesmahbub commented 3 years ago

I there way to set Tacotron 2 for Bengali language samples?

Shehan-Irteza-Pranto commented 3 years ago

@nafeesmahbub did you train it?

nafeesmahbub commented 3 years ago

Yes I trained it but the inference is blury audio. I set the checkpoints to 500 steps

On Tue, 6 Jul 2021, 11:29 pm Shehan Irteza Pranto, @.***> wrote:

@nafeesmahbub https://github.com/nafeesmahbub did you train it?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/NVIDIA/tacotron2/issues/492#issuecomment-874948628, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFFXINW3PLJI5JQ4ADHGM43TWM4OTANCNFSM46JYEZLA .

Shehan-Irteza-Pranto commented 3 years ago

@nafeesmahbub could you please suggest about the Bengali symbols which might be needed in symbols.py? And other finetunes for Bengali language?

nafeesmahbub commented 3 years ago

_characters = 'আঅইঈউঊঋএঐওঔকখগঘঙচছজঝঞটঠডঢণতথদধনপফবভমযরলশষসহড়ঢ়য়ৎংঃঁ০১২৩৪৫৬৭৮৯' I kept the configs default

uttaparsa commented 3 years ago

Did you fine-tuned the English model or trained from scratch? Does it need to be trained from scratch?