/midi-track-identification

:coffee: MIDI tracks categorization: melody, drum , bass and other

Primary LanguageJupyter Notebook

Midi-Track-Identification

Categorize tracks of midi files into four classes:

  • 0: Melody
  • 1: Drum
  • 2: Bass
  • 3: Other

There are two main goals of this task:

  • Data Cleansing:
    • The track names in midi files are usually less meaningful.
    • The program number and channel number are not in GM Midi format.
  • Separate perceptually melodic tracks based on hand-labeling data

Note that it's possible that there are more than one melodic tracks in a song.

Result

Accuracy

  • Train: 1.0
  • Test: 0.98

Dataset

  • Jazz Realbook
  • Total: 1094 songs (7:3)
    • Train: 765
    • Test: 329

Conception

  • Feature Extraction
  • Classification (Random Forest, n_estimation=100)

Figures

  • Testing Stage image

  • Feature Distribution image image image

  • For further understanding, please refer to here

Usage

python compile_dataset.py
python compile_feature.py
python analysis.py

To run testing on self-prepared data, please check out the test.py