search

craffel / mir_eval

Evaluation functions for music/audio information retrieval/signal processing algorithms.
MIT License
607 stars 114 forks source link

Move chord.evaluate_file_pair to the evaluator #62

Closed craffel closed 10 years ago

craffel commented 10 years ago

We can have a data preprocessing function (filtering and merging intervals) in .chord, like melody.to_cent_voicing: https://github.com/craffel/mir_eval/blob/master/mir_eval/melody.py#L218 but we shouldn't have a function in the task submodule which loads in data, and also we shouldn't have a function in the task submodule which runs all of the metrics - those are things which should be moved to the evaluator

craffel commented 10 years ago

Related: It looks like chord_eval prints per-file statistics, and doesn't aggregate over files. I think this is functionality that @ejhumphrey mentioned adding. Once this has been added, it can match the structure of the other evaluators -- with a evaluate function which takes in a file (or in this case a list of files), computes the scores for them, and returns a metric dictionary (here the keys would be the vocabs -- probably best to default to all vocabs) which summarizes the scores for all of the included files.

craffel commented 10 years ago

Merging with other issues into #81