DDS

This repository hosts the implementation of Differentiable Dictionary Search (DDS) method.

Install

In your Python (tested with 3.8) environment of choice:

pip install -r requirements.txt

Link the MAPS dataset:

path_to_maps=/data/MAPS # or wherever your unpacked MAPS resides
mkdir ./data
ln -s $path_to_maps ./data/MAPS

Use

• train_range.py trains a set of single-source models for specified note range
• train_ms.py trains a single multi-source model for specified note range
• test_RAND.py evaluates specified method on specified subset of MAPS/RAND
• test_MUS.py evaluates specified method on specified piece from MAPS/MUS

More detailed instructions with examples in curation.

License

All software provided in this repository is subject to the CRAPL license.