Overview

A project that trains a LSTM recurrent neural network over a dataset of MIDI files. More information can be found on the writeup about this project. This the code for 'Build an AI Composer' on Youtube

Dependencies

Numpy (http://www.numpy.org/)
Tensorflow (https://github.com/tensorflow/tensorflow)
Python Midi (https://github.com/vishnubob/python-midi.git)
Mingus (https://github.com/bspaans/python-mingus)

Use pip to install any missing dependencies

Installation (Tested on Ubuntu 16.04)

Step 1: Tensorflow version 0.8.0 must be used. On Tensorflow's download page here, scroll down to "Pip Installation". Follow the first step normally.

You will see "export TF_BINARY_URL" followed by a URL. Modify the part of the url that has "tensorflow-0.10.0", so that it will download version 0.8.0, not version 0.10.0 "tensorflow-0.8.0.

Example of the modified url, for the Python 2.7 CPU version of Tensorflow:

export TF_BINARY_URL=https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-0.8.0-cp27-none-linux_x86_64.whl

sudo pip install --upgrade $TF_BINARY_URL

Follow the third step normally to install Tensorflow.

Step 2: After installing Tensorflow, you will have to install the missing dependencies:

pip install matplotlib

sudo apt-get install python-tk

pip install numpy

Step 3:

cd ~
git clone https://github.com/vishnubob/python-midi
cd python-midi
python setup.py install

cd ~
git clone https://github.com/bspaans/python-mingus
cd python-mingus
python setup.py install

Basic Usage

mkdir data && mkdir models
run 'python main.py'. This will collect the data, create the chord mapping file in data/nottingham.pickle, and train the model
Run python rnn_sample.py --config_file new_config_file.config to generate a new MIDI song.

Give it 1-2 hours to train on your local machine, then generate the new song. You don't have to wait for it to finish, just wait until you see the 'saving model' message in terminal. In a future video, I'll talk about how to easily setup cloud GPU training. Likely using www.fomoro.com

Credits

Credit for the vast majority of code here goes to Yoav Zimmerman. I've merely created a wrapper around all of the important functions to get people started.

llSourcell/AI_Composer

ErrorLooks like something went wrong!

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Contributors 5

Languages

Error
Looks like something went wrong!

Packages