/musym-GDL

Geometric Deep Learning Applied on Symbolic Music Scores

Primary LanguagePython

musym-GDL

Geometric Deep Learning Applied on Symbolic Music Scores

Overview

Cadence and Voice Leading Detection in Symbolic Classical Music is a challenging task. This Repository provides method for Geometrical Deep Learning models and score graph modelling paradigms for applying Node Classification on Musical Score graphs for Cadence Detection.

A typical Heterogenous graph modelling of the musical score

score2graph_representation

node_attributes

Dependencies

  • pytorch 1+

  • dgl v0.6

  • pandas

Quickstart

Install requirements with pip:

pip install requirements.txt

Run a experiment :

cd src/models/rgcn-hetero
python entity_classify.py -d mps_onset --testing --gpu 0