Loading…
Results
You’re all caught up. New activity for this workspace appears here.
Loading…
MoveEnter OpenSearches page names in this workspace only
Judging
One measured analysis, presented by four AI personas, decided by people. Here is the whole path, in order.
Your recording is checked for quality and measured — pitch, timing, dynamics. Measures that do not depend on language only.
A deterministic engine turns the measurements into an AI-assisted draft. No AI language model scores anything.
Independent human judges score your entry blind — they never see your name.
The head judge reconciles the reviews and approves the result set.
A different person publishes the approved results. Only a published result is official.
If the rules allow it, you can request a review within the appeal window.
Presentation personas for parts of one AI-assisted assessment. Each is labelled AI, states its limits and explains only its evidence.
Loading the personas…
If a criterion cannot be measured reliably — no permitted reference melody, free tempo, a language without validation — it reads “Not assessed” with the reason. It is never scored as zero, and no number is invented.