How to combine twf and fft features for MLSTM-FCN model?

[georgemossad]

Hi,

I have a time series classification problem where I want to use both the time wave form (twf) and the fast Fourier transform (fft) as features for the MLSTM-FCN model. I want to concatenate them along the channel dimension and pass them to the lstm layer, and then to the fcn classifier. Is there a way to do this in tsai? Or is there a better idea to handle this problem if I want to benefit from both twf and fft features? Thank you.