groover 0.2.2

A rhythm feature extractor and classifier for MIDI files

groover 0.2.2

Installation

groover is a beat-by-beat rhythm feature clustering and token generation tool for .mid files. You can download groover using pip:

pip install groover

To check if groover is successfully installed, type python in the terminal, and do the following:

>>> from groover import RhythmClassifier
>>> type(RhythmClassifier())
<class 'groover.classifier.RhythmClassifier'>

RhythmClassifier

RhythmClassifier.__init__

Returns: None

Parameters (all optional)

• bins_per_beat: int
  • the number of quantized time units within a beat
  • is set to 24 by default
• n_beats_pitched: int
  • the number of beats that consist a rhythmic pattern for pitched instruments
  • is set to 1 by default
• n_beats_drums: int
  • the number of beats that consist a rhythmic pattern for drum instruments
  • is set to 4 by default
• drums: list
  • a list of str that specifies which types of drums to consider
  • complete mapping from drum names to MIDI pitches can be found in groover.drum_names_to_pitches
• notes_weight: callable
  • a function that takes a list of miditoolkit.midi.containers.Note as input, and outputs an numpy.ndarray with shape (len(notes),)
• similarity: callable
  • a function that takes two numpy.ndarray and output the similarity score between the two arrays

RhythmClassifier.get_pitched_dataset(self, midi_objs, pitches=None, in_four=False)

Returns: numpy.ndarray

• a dataset that contains rhythmic patterns extracted from pitched instruments of the MIDI objects
• has shape (n, m), where n is the number of patterns and m is the number of quantized time units within a pattern

Parameters

• midi_objs: list
  • a list of miditoolkit.midi.parser.MidiFile to extract rhythmic patterns from
• pitches: iterable
  • an iterable that contains the pitches to be considered
  • is set to range(128) by default
• in_four: bool
  • will only consider MIDI files that has time signatures in power of two if set to True

RhythmClassifier.get_drum_dataset(self, midi_objs, in_four=False)

Returns: numpy.ndarray

• a dataset that contains rhythmic patterns extracted from drum instruments of the MIDI objects
• has shape (n, len(self.drums), m), where n is the number of patterns and m is the number of quantized time units within a pattern

Parameters

• midi_objs: list
  • a list of miditoolkit.midi.parser.MidiFile to extract rhythmic patterns from
• in_four: bool
  • will only consider MIDI files that has time signatures in power of two if set to True

RhythmClassifier.def fit_from_midi(self, midi_objs, k_pitched=200, k_all_drums=100, k_single_drum=20, quantize=True):

Returns: None

The RhythmClassifier instance itself will be modified with updated rhythmic pattern clusters.

Parameters

• midi_objs: list
  • a list of miditoolkit.midi.parser.MidiFile to classify rhythmic patterns from
• k_pitched: int
  • the number of rhythmic pattern classes from pitched instruments
• k_all_drums: int
  • the number of rhythmic pattern classes from drum instruments
• k_single_drum: int
  • the number of rhythmic pattern classes from each drum
• quantize: bool
  • quantize drum patterns further down to 16th notes if set to True

RhythmClassifier.add_pitched_markers(self, midi_obj, pitches=None)

Returns: None

midi_obj will be modified with rhythmic pattern markers from pitched instruments in midi_obj.markers

Parameters

• midi_obj: miditoolkit.midi.parser.MidiFile
  • the MIDI file to add rhythmic pattern markers from pitched instruments to
• pitches: iterable
  • an iterable that contains the pitches to be considered
  • is set to range(128) by default

RhythmClassifier.add_composite_drum_markers(self, midi_obj, rid_empty=True)

Returns: None

midi_obj will be modified with rhythmic pattern markers from drum instruments in midi_obj.markers

Parameters

• midi_obj: miditoolkit.midi.parser.MidiFile
  • the MIDI file to add rhythmic pattern markers from drum instruments to
• rid_empty: bool
  • sections with no drums will not add markers to the MIDI files if set to True

RhythmClassifier.add_separate_drum_markers(self, midi_obj, rid_empty=True)

Returns: None

midi_obj will be modified with rhythmic pattern markers from each drum instruments in midi_obj.markers

Parameters

• midi_obj: miditoolkit.midi.parser.MidiFile
  • the MIDI file to add rhythmic pattern markers from each drum instruments to
• rid_empty: bool
  • sections with no drums will not add markers to the MIDI files if set to True