Karol Piczak

Thanks, I think that's a good point. At this moment I'm rather swamped with other projects, but when I have more resources to spare I will try to update the...

> Also, I think in the "Install Python with required packages:" section, "-add channels" should be "--add channels". Thanks for spotting that. I think I was pasting it directly, but...

That's very possible that something broke on the way, as I was checking it last time in April/May without any problems. If you still encounter difficulties with downgraded versions, at...

Great to hear that! Theano has changed status recently to no longer actively developed. Maybe that's why it was dropped from the dependency list in Keras. Thanks, I will amend...

Hmm... I will look into it. I'm not sure why newer versions would pose a problem.

Hi, I am not sure why my previous e-mail reply didn't go through, I missed that. There are no predefined train/test splits, but the dataset is split into 5 folds...

There are no predefined train/test splits, but the dataset is split into 5 folds of equal size. A common procedure is to use 4 folds for training and 1 for...

New librosa versions have changed the API, so I will have to update it accordingly. For the time being you can rename the `ref_power` parameter to `ref` for newer versions...

I will have a look at this. The Dataverse version is an older release in a compressed format (OGG Vorbis) instead of WAV (which is easier to load), with different...

Thank you. That is indeed correct. However, due to the amount of time that has passed since the original publication, I will be moving this repository to a read-only (archival)...