search

JamesOwers / midi_degradation_toolkit

A toolkit for generating datasets of midi files which have been degraded to be 'un-musical'.
MIT License
38 stars 5 forks source link

Amt #119

Closed apmcleod closed 4 years ago

apmcleod commented 4 years ago

This mainly implements the measure_errors.py script which generates a json of degradation and clean proportions given a gt and trans directory.

This fixes #30.

On a minor note, this also fixes #110.

apmcleod commented 4 years ago

This works on all of the acme degradations. However, using it, I found many errors in those files. I confirmed that those errors are present in the underlying csv, and that this measure_errors.py script outputs sensible results.

apmcleod commented 4 years ago

Combining the dataset work into amt. This is an attempt at #82 as well.

It's just a Degrader class that you can create with a config.json file or manual parameters, and then you just have to call degrader.degrade(note_df) on any input data point, and it will return a degraded_df and and degradation_id label for you to use as you wish.

apmcleod commented 4 years ago

I would say that this also fixes #82. Although it's not a Dataset, it does have the "degrade-on-the-fly" functionality.