can't pickle tensorflow

[ghost]

I have tried the main repository, and i get the error that "self" is not defined. There was an issue about this, so i used the branch from that fix. Now i am getting:
raceback (most recent call last):
File "demo_cli.py", line 93, in
mels = synthesizer.synthesize_spectrograms(texts, embeds)
File "D:\RealTimeDeepFake\voice\Real-Time-Voice-Cloning\Real-Time-Voice-Cloning\synthesizer\inference.py", line 101, in synthesize_spectrograms
[(self.checkpoint_fpath, embeddings, texts, self._seed)])[0]
File "C:\Users\fnafm\AppData\Local\Programs\Python\Python37\lib\site-packages\multiprocess\pool.py", line 276, in starmap
return self._map_async(func, iterable, starmapstar, chunksize).get()
File "C:\Users\fnafm\AppData\Local\Programs\Python\Python37\lib\site-packages\multiprocess\pool.py", line 657, in get
raise self._value
multiprocess.pool.MaybeEncodingError: Error sending result: '<multiprocess.pool.ExceptionWithTraceback object at 0x00000250FDB07DC8>'. Reason: 'TypeError("can't pickle tensorflow_core.python._tf_stack.Stack objects")'

I have tried using conda environment with python 3.7, and also, plain python, following this procedure : https://poorlydocumented.com/2019/11/installing-corentinjs-real-time-voice-cloning-project-on-windows-10-from-scratch/

My hardware really sucks and i can't run linux. I am stuck with live sessions and windows.
How can i fix the issue?

[ghost]

Hi @AntINFINAit , thank you for reporting this issue. As a workaround, remove the seed argument from this line:

Real-Time-Voice-Cloning/synthesizer/inference.py

Line 109 in 8f71d67

model = Tacotron2(checkpoint_fpath, hparams, seed=self._seed)

That is, change it to:

model = Tacotron2(checkpoint_fpath, hparams) 

You will be not have the ability to force deterministic synthesis, but at least it should run. If you need that feature, try the 423_add_cpu_mode branch of my fork, and change the --low_mem argument to --cpu.

[ghost]

The issue is that "self" does not appear to be referenced in this python. That is why i used your other branch. How can i fix that error?

[ghost]

I should have been a little more clear, you should be able to make the above change on Corentin's master and it should also resolve the "self" issue by not referencing it.

Since you're using my fork as the starting point, try replacing seed=self._seed with seed=None in this line:
https://github.com/blue-fish/Real-Time-Voice-Cloning/blob/8e4a0565544d0ae6ee3d22518961a28ed88eba5e/synthesizer/inference.py#L100-L101

If that doesn't work then try the CPU mode since that bypasses the low_mem code that is giving you issues.

[ghost]

@AntINFINAit Are you still experiencing this problem?