Skip to content

Instantly share code, notes, and snippets.

@tikhonova
Created April 16, 2023 22:30
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save tikhonova/82c7a6a771dadec3f3e18903e1b9dcf2 to your computer and use it in GitHub Desktop.
Save tikhonova/82c7a6a771dadec3f3e18903e1b9dcf2 to your computer and use it in GitHub Desktop.
''' ffmpeg not found when using pydub utils
___
If using Win, need to download from the official website and add to path, then reload git bash.
https://github.com/jiaaro/pydub/issues/348
'''
''' AssertionError: Distributed mode requires CUDA
___
a MUST-read to confirm that both GPU and drivers support the CUDA version you've installed (or about to install):
https://stackoverflow.com/questions/60987997/why-torch-cuda-is-available-returns-false-even-after-installing-pytorch-with
Tacotron2 script was not seeing my cuda. I have a GPU RTX 2070 and CUDA toolkit 11.7.
Checks:
>> nvcc -V (to get detail re nvcc: NVIDIA (R) Cuda compiler driver)
>> nvidia-smi (to obtain GPU info)
>> conda list cudatoolkit (to see package info confirming installation)
>> python -c "import torch; print(torch.cuda.is_available())" (it printed False for me, which means PyTorch was not compiled with CUDA support,
so I had to reinstall it with the cuda flag following the steps from the next assertion error described below.)
In the end, it 'Successfully installed +cu117'.
Note that on Windows you must also specify 'gloo' backend in hparams.py as Windows doesn't support nccl distributed computing as of the time of writing.
'''
# via https://github.com/tikhonova/what_would_alan_watts_say/blob/master/speech_synthesis/debugging_errors.py
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment