No Arg Named axis for torch.cat

[Aloento]

https://github.com/anonymous-pits/pits/blob/57ca2ef3778396f0f15cd5f186a30925a5e8795e/models.py#L705

https://pytorch.org/docs/stable/generated/torch.cat.html#:~:text=the%20cat%20dimension.-,dim,-(int%2C

[anonymous-pits]

Hi Alento,

It is true that axis is not written on official docs, but it is working code.

I did not notice it, because I brought that part from official avocodo implementation and working well.

I will fix it in next monday.

[Aloento]

for newer torch version

in spectrogram_torch -> torch.stft and mel_spectrogram_torch

RuntimeError: stft requires the return_complex parameter be given for real inputs, and will further require that return_complex=True in a future PyTorch release.

Add return_complex=False or .view_as_real()

But

https://pytorch.org/docs/stable/generated/torch.istft.html#:~:text=datatype%20inputs%20are-,no%20longer%20supported,-.%20Input%20must%20now

So wait

https://github.com/mindslab-ai/phaseaug/issues/8

[Aloento]

for new version of librosa, in mel_processing.py

TypeError: mel() takes 0 positional arguments but 5 were given

rewrite to:

librosa_mel_fn(sr=sampling_rate, n_fft=n_fft, n_mels=num_mels, fmin=fmin, fmax=fmax)

[Aloento]

for newer scipy, in pqmf.py

from scipy.signal.windows import kaiser

[anonymous-pits]

Modifying our current implementation to fit the version released last week is challenging.

For now, if you want to use our implementation, please use a Torch version similar to the one specified in the Dockerfile (around Torch 1.9.0).

[Aloento]

It's working with pytorch 2