crim_intervals 0.2.1

A python based, interactive music similarity engine

Building a music similarity engine based on intervals

Find the project on Github and PyPI

Github PyPI

Project Goals:

• Load mei file(s) from url or path
• Carve out desired portion of piece/corpus
• Create a sequence of interval objects from the score
• Set filters on parts, measures, ids
• Export ema
• Analyze similarity of two objects

Assisted Usage

For a guided way to get results for the basic intended usages of the project, simply enter:

assisted_interface()

wherever you are writing your code.

Sample Usage

• One piece at a time

piece1 = ScoreBase('https://crimproject.org/mei/CRIM_Model_0008.mei')
piece2 = ScoreBase('https://sameplemeifile.mei')
vectors1 = IntervalBase(piece1.note_list)
vectors2 = IntervalBase(piece2.note_list_all_parts(1, 20))
patterns = into_patterns([vector1.generic_intervals, vector.generic_intervals], 5)
exact_matches = find_exact_matches(patterns, 10)

• Loading in a Corpus

corpus = CorpusBase(['https://sameplemeifile1.mei', 'https://sameplemeifile.mei'],[/sample/path/to/mei/file, /sameple/path/two])
vectors = IntervalBase(corpus.note_list)
patterns = into_patterns([vectors.semitone_intervals], 5)
close_matches = find_close_matches(patterns, 10, 1)

• Outputting relevant information
  • Printing out matches information

  ...
  exact_matches = find_exact_matches(patterns, 10)
  for item in exact_matches:
      item.print_exact_matches()
  close_matches = find_close_matches(patterns, 10, 1)
  for item in close_matches:
      item.print_close_matches()
  

  • Similarity scores

  piece1 = ScoreBase('https://crimproject.org/mei/CRIM_Model_0008.mei')
  piece2 = ScoreBase('https://sameplemeifile.mei')
  print(similarity_score(piece1.note_list, piece2.note_list, 5))
  

  • Outputting match information to csv (includes ema, mei slices)

  export_to_csv(exact_matches)
  

  • Find occurences of a motif

  find_motif(CorpusBase object, [motif], generic intervals boolean)
  

Usage Flow ~~~

• Load in files with either ScoreBase or CorpusBase

  ScoreBase(url)
  CorpusBase([url1, url2, ...], [filepath1, filepath2, ...])
  

  • Search for a motif:

  find_motif(corpus: CorpusBase, [motif], generic intervals: boolean)
  

• Create desired note list for use in IntervalBase
  • Options using CorpusBase:

    piece.note_list
    

  • Options using ScoreBase:

    piece.note_list
    piece.note_list_down_beats
    piece.note_list_selected_beats([beats])
    piece.note_list_all_parts(starting_measure, number_of_measures_after)
    piece.note_list_single_part(part_number, starting_measure, number_of_measures_after)
    piece.note_list_by_offset([offsets])
    

• At this point similarity scores can be shown
  • size of pattern indicates how many notes in a row need to follow the same rhythmic pattern to be considered a match

similarity_score(first piece note list, second piece note list, size of pattern)

• Decide between semitone intervals and generic intervals

vectors.generic_intervals
vectors.semitone_intervals

• Construct patterns from note_list (more options for pattern construction forthcoming)

into_patterns([list of piece note lists], size of pattern)

• Decide which type of matches to find
  • These will not work if you don't send it the return value from into_patterns

find_exact_matches(return value from into_patterns, minimum matches needed to be displayed)
find_close_matches(return value from into_patterns, minimum matches needed to be displayed, threshold)

• Returns a list of matches for easy analysis, printing shown above
• Export to csv in current working directory:

export_to_csv(exact_matches)

• Run desired analysis with your own python code, print out results, etc.