I don't think so. I only do Legacy Tacotron since I upload my models on FakeYou as well.

Once the model's finished, I just go ahead with synth tests.

can anyone send me working tacotron 2 co;ab notebook link

@everyone

Hello, I'm trying to use the Uberduck Pipeline HiFi-GAN Training notebook but I am left with this error on the 'Generate Mel Spectrograms' cell:

Not using any style tokens

---------------------------------------------------------------------------

RuntimeError                              Traceback (most recent call last)

<ipython-input-10-a8fe2391cead> in <module>
     56 
     57 device = "cuda:0"
---> 58 taco.from_pretrained(warm_start_path="TTSModel.pt", device="cpu")
     59 if device != "cpu":
     60     taco.cuda()

1 frames

/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py in load_state_dict(self, state_dict, strict)
   1405         if len(error_msgs) > 0:
   1406             raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
-> 1407                                self.__class__.__name__, "\n\t".join(error_msgs)))
   1408         return _IncompatibleKeys(missing_keys, unexpected_keys)
   1409 

RuntimeError: Error(s) in loading state_dict for Tacotron2:
    size mismatch for embedding.weight: copying a param with shape torch.Size([148, 512]) from checkpoint, the shape in current model is torch.Size([160, 512]).

Also I do not know where to upload the wavs

Using the Russian symbol set

they must be in a zip file

in the same directory as your list

ok, what name?

oh wait

im dumb

ok it still doesn't work

WAIT NOOOO DID I CORRUPT IT

hmm im looking at the error and it seems to be an issue with the model itself

are you using finalized

i am using legacy tacotron2

oh uh

that error should like NEVER exist with legacy

uhhhh

I think its corrupted. I need to retrain i think

that’s probably the most likely solution but i would test just synthesis to corroborate

this is synthesis

ohhhh

then im at a loss

yeah its corrupted

i'll retrain, it was terrible anyway (probably because i used a Russian transcript just as an experiment)

why does this uh happen

there is no such thing as wav.wav

why does it say wav.wav.