How does music evolve?
SeraWorld
SeraWorld studies musical states, their relationships, and the consequences of change.
Inside SeraWorldMUSIC / RESEARCH / PRACTICEMusic,
Music,
intelligence
& form.
Exploring how music evolves.
And how we might create with it.
CALLING / 2026For small chamber ensemble
Musical intelligence · Human–AI collaborationSCROLL TO DISCOVER
01 / RESEARCHA connected research practice
Two questions.
One musical world.
My research explores systems that model musical evolution and collaborate reliably with people across long creative processes.
How should AI act on music?
SeraForge
SeraForge studies editing tools, project memory, and reversible workflows that preserve a composer's intent.
Inside SeraForgeA shared foundation: Sera Core
Authoritative score state, musical operations, and reversible patches.
02 / SELECTED WORKSystems for thinking through music
Ideas, made
tangible.
Research takes shape through instruments, software, and experiments. A selection from the Sera ecosystem.
COMPOSITION / 2026
CALLING
For small chamber ensemble
03 / SOFTWARE IN ACTIONSeraEdit · Interactive replay
A musical idea.
A traceable change.
Follow a real edit from an instruction to a validated score revision. Explore six recorded examples from the offline software demonstration.
Recorded software execution. The walkthrough replays saved results; it does not call an AI model.
Download evidence ↗Precise musical scope
Address measures, staves, voices and individual events. Bind every proposal to the score it was created for.
SERAEDIT / SERA COREIntent that persists
Retain project memory and lock musical properties, such as a motif's rhythm, across an editing session.
SERAFORGEChanges you can inspect
Review differences, validation results and candidate alternatives before accepting a reversible change.
SERAFORGE / SERA COREMusical consequences
Study how timing, voice and context affect the predicted consequences of an edit, including when to abstain.
SERAWORLD / IN DEVELOPMENT04 / RESEARCH & RELEASESEvidence, questions, next steps
From exploration
to evidence.
A released research-software foundation, a manuscript in peer review, and two active lines of investigation.
SOFTWARE PUBLICATION TRACK
SoftwareX
Under ReviewLast documented status
SeraEdit 1.0.0 · Yuan Gao
Reliable Language-Guided MusicXML Editing through Structured Score Patches
A local-first editing system that turns language-guided proposals into bounded, validated changes, with explicit protection for the rest of the score.
20Synthetic benchmark scores
120Editing tasks in the release
6 / 6Offline demo cases passed
Rerun September 15, 2026
Rerun September 15, 2026
- Software archive recorded · v1.0.0
- Submitted to SoftwareX
- Under Review recorded
Manuscript SOFTX-D-26-01135. Peer review is ongoing in the latest documented record; acceptance has not been reported. The offline demonstration checks software behavior and score round trips.
Correct steps.
Coherent journeys?
In a deterministic development probe, format repair improved local joint success from 80% to 84%. For transposition, intended-trajectory retention remained at 30% in both conditions.
Local success · format repair84%
Transposition trajectory retention30%
The next challenge is preserving the cumulative musical goal across many edits.
Study scope & evidence
September 12, 2026 · 18 sessions × 100 steps; 3 synthetic layouts from one generator family. Each condition used two previews per turn. No live LLM comparison or human study. Safety gates still accepted some valid but wrong-value edits.
Knowing when
a prediction is uncertain.
The current prototype models the consequences of changes with event timing and voice context. Work-separated calibration tests examine how much of that prediction can be trusted.
Musical state→Proposed action→Consequence + uncertainty
Broad uncertainty intervals still limit useful coverage. The strict reliability gate has not passed.
Study scope & evidence
September 10, 2026 · 10 new calibration works and 4 development-validation works passed source admission. Independence refers to this predictor's training and earlier calibration, not all historical project data. Multi-step planning and musical usefulness remain open.
ONE ECOSYSTEM. DISTINCT QUESTIONS.
SeraWorld models musical evolution. SeraForge studies reliable human–AI musical action. Both share authoritative score infrastructure through Sera Core; Forge can run independently of World.
05 / RESEARCH NOTESA record of work in progress
An evolving
practice.
Where the work stands, what remains open, and the questions that guide the next experiment.
Research snapshot · September 15, 2026. Milestones describe development progress, not published research findings.
06 / ABOUT
Yuan Gao.
AT THE INTERSECTION OF MUSIC & AII am interested in the space between musical imagination and intelligent systems.
Through the Sera research ecosystem, I explore how computational systems can understand musical change and support composers while keeping creative decisions in human hands.
This site brings together my research in progress, the tools that emerge from it, and a growing archive of musical work.
Follow the research07 / CONTACTResearch · Software · Composition
Let's start
a conversation.
For research conversations, software questions, or musical collaborations, you can reach me at any of these addresses.