Concatenate Error

[Kristopher-Chen]

Hi, have you ever met this problem?

in "training_tf2/lpcnet.py", line 246, in new_lpcnet_model,
ValueError: A Concatenate layer requires inputs with matching shapes except for the concat axis. Got inputs shapes: [(128, None, 20), (None, None, 64)]

Thank you!

[norihirosunada]

Any answer to this issue?

I got same problem while trying to train my own model.
I think I've followed right steps.
I have used src/concat.sh to get input.s16 and used dump_data to get features.f32 and data.s16.
And runned training_tf2/train_lpcnet.py.
Anything else to concern about?

Thanks!

[exercise-book-yq]

No one can answer to this issue?

[norihirosunada]

Version of Python might be the answer to this issue.
I've rebuilt my environment and this error got away.
I was using Python 3.6.8 and got this error, but rebuilt the environment with 3.8 and concatenate error didn't appear.
Hope this will help you.