THE PREMISE
Music has geometry. We construct within it.
Pythagoras described music as ratios and Kepler connected it to planetary motion. Composers operate within a consistent space defined by the relationships among notes, harmonies, rhythms, and arrangements. These elements decide whether music is coherent, resolved, or unsettled.
SHE works inside this space by directly using these relationships. It does not depend on listening to existing music.
ORIGIN
Where It Started
The Structured Harmony Engine (SHE) began with a long-running fascination: music theory, experimental composition, and procedural art. That work ran alongside twenty years building metadata systems, machine learning platforms, and recommendation infrastructure for music and video streaming services, which made two pressures plain.
More music, less provenance
Generated music is growing in volume, and it is getting harder to trace where a piece came from or how good its data is.
Decisions nobody can see
Closed models do not show how a piece was made, so nobody can manage those decisions precisely.
The question became: could a system understand music deeply enough to compose it, mix it coherently, and stay original?
The answer was a lightweight engine with endless possibilities. Its material is symbolic music data, music captured as notes and instructions rather than a finished recording, and it composes from the relationships between those notes directly. That is the artistic bet behind SHE: that the rules of music, written down with care, are enough to make new music without end.
SHE was not trained on anyone's audio and uses no generative AI or large language models. We built the engine ourselves and control every part of it.
THE TEAM
The People Behind It
The team is using their experience with music, audio engineering, and taxonomy design to build a data-driven, rule-based system for music. What started as an experiment has become a stable platform for music creation, education, and research.
Taylor Kirch
Co-Founder & CEO
- Netflix
- Pandora/SiriusXM
- MS Information Science, UIUC
Taylor Kirch is a product leader and contemporary artist. As SHE's lead product designer and music technologist, she drove the engine from concept to production, designing its architecture and building the taxonomies of genre, instrument, mood, and rhythm.
She brings over a decade of experience at the intersection of music, metadata, and machine intelligence.
She's a classically trained singer, with years of experience in chamber, choir, a cappella, and opera music.
Tre Tomaszewski
Co-Founder & CTO
- PhD research, UIUC
- Bioinformatics
- Audio engineering
Tre Tomaszewski is a researcher and multidisciplinary technologist whose PhD research sits at the crossroads of cognitive psychology and AI: information networks, AI-assisted cognition, agent social interaction, and machine cognitive psychology.
Earlier, as a bioinformatics graduate student, he published research on genetic mutation detection in COVID-19 that seeded further work on the same data infrastructure.
Outside academia, he has worked across music and gaming as a designer and programmer. He has experience in original music composition and audio engineering.
Music has geometry. We construct within it.