Loading…
Loading…
Judging transparency
How scores are produced, what the AI personas can and cannot measure, and who decides. The personas present one set of automated measurements; people decide every result.
The pipeline
Each persona presents a slice of the same automated analysis. None of them scores on its own, none of them hears your performance as a listener would, and every number they present is reviewed by a human judge before it counts.
One continuous take, exactly as you recorded it.
Pitch, timing and phrasing measured where possible — “Not assessed” where not.
MELODY, RHYTHM, HARMONY and STAR present slices of the same measurements.
Judges score every criterion, weigh the evidence and may overrule it with a written reason.
Published only after a head judge approves it.
The AI judges
Four ways of presenting one set of automated measurements — never the final word. People decide every result.
AI persona · syntheticPersona copy pending approval
What it presents
Pitch & melody
Pitch & melodic analysis
Its limits
Measures it presents
Four presentation personas over one analysis pipeline and human review — not four independent AI judges, and none of them decides anything.
Languages
What the automated analysis can do differs by language. We publish five capabilities for every language, separately, and mark anything unvalidated as “Not assessed”.
Can you enter singing in it? For every listed language: yes.
Is this website translated? Complete, partial or English fallback.
Not available for most languages yet; shown as “Not assessed” when missing.
Understanding of the words you sing — validation pending.
Profiles such as Carnatic or maqam are pending — no Western default is applied.
English · English
العربية · Arabic
தமிழ் · Tamil
Capability values are provisional. Tagline translations are samples, not yet reviewed by native-speaker editors.
Lines the AI doesn’t cross
No result is official until human judges approve it.
Anything it can’t measure for your language or tradition is “Not assessed”, never a number.
Every persona is an abstract form, labelled as AI wherever it appears.
Audience reactions are never used in scoring.
Human judges review every automated finding and decide every result.