Code needed for building the data and learner for predicting chiptunes using MIDI representations
Python Shell
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
GamePost
contrib
data
docs
generated
log
survey
utilities
.gitignore
.gitmodules
README.md
TODO.txt
bayes_net.py
create_training_data.sh
hmm.py
max_min.py
predict_batch.py
predict_songs.py
random_noise.py
random_songs.py
song_collections.py
stats.py
test.py
test_predict_bayes.py
test_predict_hmm.py
test_random_songs.py
utils.py

README.md

nes-machine-learning

Code needed for building the data and learner for predicting chiptunes using MIDI representations

Files

data/
    raw/*
        - All of the MIDI files that have been generated and collected so
          far
    all_songs/*
        - A subset of the *raw* directory that only includes songs that
          have been manually labeled as "songs", as opposed to empty
          tracks, sound effects or other background songs.
    training_songs/*
        - A subset of the *all_songs* directory, approx 2/3 of it,
          containing songs that should be used to generate and train
          the algorithm with.
    testing_songs/*
        - A subset of the *all_songs* directory, containing the approx
          1/3 of the songs that are not in the *training_songs* directory,
          that should be used for testing the accuracy of the algorithm
    training_counts.sqlite3
        - The generated sqlite3 database of transition counts (generated
          with the *training.py* file) based off the *training_songs*
          directory

Installing

install python-midi by running:

  1. cd contrib/python-midi
  2. python setup.py install

install MIDIUtil by running:

  1. cd contrib/MIDIUtil
  2. python setup.py install

You'll also want to generate some training counts. You can do so with:

  1. python training.py

Related Software

  • nes2midi Windows tool for converting NSF files to MIDI. The original software is in Japanese, but there is thankfully an English translation too!
  • WINE Windows DLL implementation used for running the above software on OSX and Linux. OSX instructions are available on the WINE site.