AUDIO ML · archived
SongDNA
independent experiment in interpretable music embeddings.
What I built
an archived research prototype that represented production style, structure, and musical relationships in a 32-dimensional embedding. it is not affiliated with spotify's later product of the same name, and no causal relationship is claimed.
the engine
Represent musical structure with interpretable dimensions.
The prototype explored a 32-dimensional representation of production style, structure, and musical relationships for recommendation and research experiments.
- 32-dimensional interpretable embeddings per track
- Designed for recommendation and research experiments
- Archived prototype; no public implementation is currently linked
scope
Separate the prototype from later products with the same name.
Spotify later released an unrelated feature called SongDNA. The shared name is noted only to prevent confusion; this portfolio does not claim contribution, adoption, licensing, or influence.
- No Spotify affiliation claimed
- No public repository currently linked
- Kept as an archived research direction rather than a deployed product
Artifacts and limits
Project images and measurements are shown as project-produced material, not independent validation.
This page intentionally avoids architecture and performance claims that cannot be inspected from a repository, demo, or report.
archived
Last content review: July 2026. Scope is limited to the material linked on this page.
Return to projects