Please first create an environment using conda or venv with python3.7. Then install the dependencies as in requirements.txt
.
To run the DBM
method,
- Prepare midi pieces to be used for building the database in the
data
folder. - Prepare also the piece to be reduced.
- go to
DBM/build_up.ipynb
and modify the directories, then run the code!
(Note: If you need to run the training code, a powerful GPU is recommended.)
- Prepare midi used for pretraining
- Follow
MidiBERT/Pretraining.ipynb
and modify relevant arguments
Please follow MidiBERT/MBNR.ipynb
for more instructions regarding training and inferencing.
- Prepare data following
MidiBERT/skinlineTokenize.ipynb
. - Run
python MidiBERT/CP/main.py --mode seq2seq
and add other arguments as required.
All objective evaluation codes are included within the eval
folder.
- Make any necessary directory changes in
eval/eval.py
- Run the code. You will get a pickle file which contains a directionary of the