[Bug] Use Spawn Instead of Fork on Windows

[Sniper296]

Describe the bug Attempting to train a tacotron model on Windows (Anaconda) fails immediately due to multiprocessing configured to use fork. Multiprocessing#Contexts and start methods

To Reproduce Steps to reproduce the behavior:

1. conda env create -f environment.yml environment.yml
2. conda activate tts
3. python tacotron_train.py tacotron_train.py

   Traceback (most recent call last):
   File "tacotron_train.py", line 5, in <module>
   from TTS.trainer import init_training, Trainer, TrainingArgs
   File "C:\Users\Adam\.conda\envs\tts\lib\site-packages\TTS\trainer.py", line 46, in <module>
   multiprocessing.set_start_method("fork")
   File "C:\Users\Adam\.conda\envs\tts\lib\multiprocessing\context.py", line 247, in set_start_method
   self._actual_context = self.get_context(method)
   File "C:\Users\Adam\.conda\envs\tts\lib\multiprocessing\context.py", line 239, in get_context
   return super().get_context(method)
   File "C:\Users\Adam\.conda\envs\tts\lib\multiprocessing\context.py", line 193, in get_context
   raise ValueError('cannot find context for %r' % method) from None
   ValueError: cannot find context for 'fork'

Expected behavior Training to start.

Environment (please complete the following information):

• OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Windows 10 64-bit 19043.1237
• PyTorch or TensorFlow version (use command below): 1.9.1
• Python version: 3.8.12
• CUDA/cuDNN version: CUDA 10.2 cuDNN 7.0
• GPU model and memory: GeForce GTX 1070 Ti 8GiB driver 471.96
• Exact command to reproduce: See steps

[erogol]

I don't have a Win system. We should wait for someone to send a ✨PR✨ for a fix.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. You might also look our discussion channels.

[bitnom]

This continues to occur on windows after installing via pip.

[bitnom]

Temporary fix until PR is accepted:

1. Clone TTS repo.
2. In requirements.txt replace trainer with trainer4win
3. Build TTS package and install via pip.