How to load local MIDI file dataset?

[cifkao]

Is there currently an easy way to load a local MIDI dataset from a directory? This seems like one of the most basic use cases for this dataset machinery. For example, I would like to use the Lakh MIDI dataset which I have, without downloading it again. Should a MIDIFolderDataset be added? Or should support for MIDI be added to MusicDataset?

[vandrw]

I have tried basing a class on muspy.FolderDataset, which seems to work fine for me.

class MidiFolderDataset(muspy.FolderDataset):
    _extension = "midi" # Change this based on your extension.

    def read(self, filename: Union[str, Path]) -> Music:
        """Read a file into a Music object."""
        return read_midi(self.root / filename)

You can now import the dataset:

music_dataset = MidiFolderDataset(data_dir, convert=True) #You can also avoid the convert parameter to enable on_the_fly conversion.