CHORDS + MELODY → ANALYSIS
The functions of notes,
one relationship at a time.
Explore a tune, follow its proof and find the same functional chains across the repertoire.
Deterministic analysis.
Observable counts.
Generation to explore.
FROM SOUND TO FUNCTION
The event sequence
A new event begins when the note, the chord, or both change; a repeated note attack also begins an event. A held note may span several events. Labels ignore octave.
ORDERED COMPOSITION
Inside the proof tree
Select a fragment to inspect its composition.
Excerpt LaTeX source bussproofs
Inspect role assignments and their evidence
| Event | Measure | Note / chord | Factor / role | Evidence |
|---|
Source and import limitations
SAME FUNCTIONS, DIFFERENT TUNES
Chains of 2 to 8 events
Exact matching of factors and roles, using overlapping windows within proven fragments. Rests and unanalysed transitions break the count. Synthetic examples are excluded.
| Chain | Occurrences | Tunes | % of windows | Explore |
|---|
FROM FREQUENCIES TO SAMPLING
Which function could come next?
An empirical distribution of observed continuations. If a context has no observations, progressively shorter suffixes are used. The sample is a functional label; realising it as notes and chords is the next stage.
Temperature 1 preserves the estimated probabilities; higher values give rare outcomes more weight. A fixed seed makes sampling reproducible.
SAMPLED LABEL
Random selection happens here.
Tune analysis remains deterministic.
Import and backend connection
The interface connects to a separate Python analysis backend.
Web import requires the backend administrator key. Command-line import is also available. Structural import errors exclude a tune from corpus counts.