PROJECT / Machine learning / CSE425 / Completed

Music Context Understanding

Connecting what music sounds like with what it means. Relation-aware audio graphs meet DistilBERT, with leakage-controlled evaluation.

PythonPyTorchGNNDistilBERT

GNN-BERT Music Context cover with headphones, an audio waveform and a connected graph. — Project cover artwork supplied by Tanzimul Karim.

CONCEPTUAL SYSTEM OVERVIEW / NOT AN APPLICATION SCREENSHOT

Audio representations

Relation-aware graph network

DistilBERT language representations

Comparative evaluation

01 / CONTEXT

From an idea to an implementation.

Connecting what music sounds like with what it means. Relation-aware audio graphs meet DistilBERT, with leakage-controlled evaluation.

This CSE425 coursework brings together Python, PyTorch, GNN, DistilBERT.

02 / ENGINEERING

Under the surface.

Conceptual overview of the documented research components; not an exact training graph. Dataset, hyperparameters, metrics and implementation decisions should be read in the linked source report.

Architecture view / inspect components

Component 1Audio representations
Component 2Relation-aware graph network
Component 3DistilBERT language representations
Component 4Comparative evaluation

This view organizes the known components. Consult the repository for exact interfaces, data shapes and implementation.

03 / EVIDENCE & LIMITATIONS

Source first. Claims second.

Verified results, benchmarks and a detailed engineering retrospective have not been supplied for this portfolio. No numerical outcomes are claimed. The source repository is available for technical review.

Explore source code

Next: Global River Database (GRD)