Shape doesn't match when inferencing Wavegram_Logmel_Cnn14 model

[LllHan]

Hi,
I used my own pre-trained Wavegram_Logmel_Cnn14 model to inference.py file with the following error:

Traceback (most recent call last):
  File "test.py", line 948, in <module>
    audio_tagging(args)
  File "test.py", line 892, in audio_tagging
    batch_output_dict = model(waveform, None)
  File "/root/miniconda3/envs/sed/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/root/miniconda3/envs/sed/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 159, in forward
    return self.module(*inputs[0], **kwargs[0])
  File "/root/miniconda3/envs/sed/lib/python3.7/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "test.py", line 563, in forward
    x = torch.cat((x, a1), dim=1)
RuntimeError: Sizes of tensors must match except in dimension 2. Got 68 and 69 (The offending index is 0)

[lbroad-tm]

This happens because or round/flooring that occurs in when creating the wavegrams and spectrograms, try interpolating your inputs onto different grids - for instance I got an error when my input was (60, 220500), but not for (60, 180000).