How to deal with the datasets?

[DeepStillWater]

Hello,I want to run your code,But I dont know how to handle the dataset ,and the file of .hdf5 I also dont know how to do it?

[topel]

Hi,
create_hdf5_ff1010bird_public.py is used to create a HDF5 file to store your dataset split into 3 subsets: 'Train', 'Valid', 'Test'. Before using it, you need to generate files containing the features. In my case, I used log-filter-bank coefficients generated from the audio files with the Matlab file MIR_extract_logSpectrumBands.m
Hope that helps!

[DeepStillWater]

Thank to reply me. I find the file create_hdf5_ff1010bird_public.py used "fbankdir + '/' + nom + '.npy",but I did not find anywhere,canyou share it with me?thanks!