danielrothmann/temporal_audio_embeddings

An autoencoder for creating embeddings of immediate audio spectra with temporal context.

temporal_audio_embeddings's Issues

Features shape problems during training

Hello @danielrothmann,
Thanks a lot for this work, it sounds very interesting and promising from your detailed articles on medium.

I was trying myself the main_listener notebook on a subset of the 8k UrbanSounds.
I have this error during the fit of the autoencoder (epoch 1, step 9):
ValueError: cannot reshape array of size 450000 into shape (563,8,100)
the batch_features shape is (1000, 8, 100) and my keras version is 2.2.4.
Any idea on what is going wrong?

Thanks a lot in advance for your help!