The librosa.util.frame function used for preprocessing the audio gives non-contiguous segments of audio. That is not what is desired in the model, or is it?
I would like to ask the author, won't the correct way to segment the audio would be to pass axis=0 parameter?
The librosa.util.frame function used for preprocessing the audio gives non-contiguous segments of audio. That is not what is desired in the model, or is it? I would like to ask the author, won't the correct way to segment the audio would be to pass
axis=0parameter?