paveyry / LyreLand

Music composition AI based on Deep Neural Networks (LSTMs).
http://lyreland.veyry.fr
MIT License
8 stars 6 forks source link

Build Status

LyreLand

Status

This project is not maintained anymore. Please use at your own risks.

Overview

LyreLand is a music composition AI that uses deep artificial neural networks (LSTM) to generate music. It also contains a package used for more traditional music analysis of MIDI files (harmonic, melodic, and rhythmic analysis) that can be used to generate music procedurally using Markov Chains trained using the analysis of a dataset.

Dependencies

LyreLand is developed using JAVA 8. It uses the jMusic library to generate MIDI scores and Fluidsynth to produce WAV files using a SF2 GM soundfont. It also uses deeplearning4j for the LSTM implementation and Spark for the web interface and website.

File storage

This repository uses git lfs to store binary files. You need to install git-lfs to work on this repository.

Building

The project dependencies are handles using maven. We use the plugin appassembler to provide simple run scripts. To build the project, run the following command:

mvn package appassembler:assemble

Running

Once the project is built, you can run the script with the following command:

sh target/appassembler/bin/app

If you want to run the servlet, use the following command:

sh target/appasembler/bin/webserv