This repository exists to contain the training process of several drum sequencer features.
- GrooVAE - learns to add expression to quantized drum rhythms
- Syncopate - learns to generate syncopated variations of drum rhythms
- Fills - the network can be conditioned on beat type to selectively generate fills or beats
Create the conda environment:
conda env update -f environment.yml
This will create an environment named dsvae. Activate the
environment as follows:
conda activate dsvae
Inside the dsvae conda environment, the dsvae
package can be installed in editable
mode
as follows. From the same directory as this README, issue the
following command:
pip install -e .
One option for maintaining sensitive data is to store values as environment variables in a .env file and use the direnv program to source this file each time you enter the repo.
We provide a sample file example.env, which stores environment variables expected in different parts of the code base. Simply copy this file to .env and fill in expected values.
You will also need to install direnv for your system.
To run the tests, run:
pytest