This repository includes the project for the third homework of the course "Deep Learning for Music Analysis and Generation" lectured by Prof. Yang at the National Taiwan University. The main goals of this work is to train a midi generator model on the pop1k7 dataset. During inference, the model will randomly generate midi and audio pairs. Most of the codes are based on the sample tutorials provided by the TAs of the course.
pip install -r requirements.txt- Please download the model weights from Google Drive: Link
- Place the checkpoint in the checkpoint folder under a specific name. e.g. checkpoint/model_v1/weight.pkl
- Inference the model with the following command:
python -m main \
--model_ver=v2 \
--mode=test \
--ckp_path=checkpoints/model_v1/weight.pkl \
--num_samples=20 \
--temperature=3.0 \
--sample_mode="topk" \
--topk=3 \
--out_prefix="temp3d0_topk3"- The command in the example above will locate the generated midi and audios in the folder 'results/model_v1/temp3d0_topk3/'.