Awesome

LSTMetallica

A LSTM network that learns from the drum tracks of Metallica and generates new tracks.

Prequisite

• Python 2.7. Some of the codes would mis-behave with Python 3.
• keras, a deeplearning framework
• python-midi, to get midi file
• numpy, probably you already have it.

Usage

• Clone the repo
• $ python main_lstM_etallica.py to get generated drum track in text file
• text->midi: $ python main_post_process.py - this is when you need python-midi
• Use this text file, an aggregated-and-encoded text file for Metallica's drum tracks, to do something more
• This folder contains the original drum midi tracks.

External links

• Details in my blog post
• Results on soundcloud and +1 more result
• Similar work on jazz chord progression: github repository, blog post, soundcloud

Citation

Text-based LSTM networks for Automatic Music Composition, Keunwoo Choi, George Fazekas, Mark Sandler, 1st Conference on Computer Simulation of Musical Creativity, Huddersfield, UK, 2016, arXiv, pdf, bib