Kid mode 🧒

The book said "at bar 13 the singer sings these three notes." The computer looked at bar 13 of the song and did not find those notes. This page tries shifting the guess by one bar, two bars, etc, to see if the book meant a nearby bar. If none of the shifts make it right, the case gets sent to a human.

Believeyoume validation · P5 of P1–P5 · 2026-07-07

Auto-fix bar-offset — Asleep In The Trunk

Takes every P4 contradicted verdict and probes bar shifts in ±10. If a shift flips the verdict to supported, it becomes an accepted fix. Otherwise the claim is flagged for manual review with the closest partial matches surfaced.

Recommendation summary

2
Contradicted probed
0
Apply shift
2
Manual review
0
Keep as-is

Offset strip legend: supported partial contradicted untestable

Per-claim probe results

manual_review vocal_sequence bar 13: B♭3 → C4 → C♯4 (card 2)

Original P4 verdict: contradicted — only 0/3 beats match. Observed: {"vocal_pcs": [9]}

Offset probe ±10:

-10-9-8-7-6-5-4-3-2-1+0+1+2+3+4+5+6+7+8+9+10

No clean shift found in ±10. Highest partial matches: offset -5 (2/3 beats match), offset +3 (2/3 beats match), offset +4 (2/3 beats match)

manual_review vocal_sequence Bar 16: B3 → C4 → B♭3 → A3 (card 2)

Original P4 verdict: contradicted — only 1/4 beats match. Observed: {"vocal_pcs": [1, 0, 0]}

Offset probe ±10:

-10-9-8-7-6-5-4-3-2-1+0+1+2+3+4+5+6+7+8+9+10

No clean shift found in ±10. Highest partial matches: offset +7 (3/4 beats match), offset -8 (2/4 beats match), offset -7 (2/4 beats match)

Honest read

Both Asleep vocal-sequence contradictions probe to manual_review, not a clean shift. The closest partial matches sit at offsets of five or more bars, which is too big a jump to accept blindly. The likelier root cause is that the walkthrough MIDI does not encode the same vocal content as the full-song MIDI the model-rebuild describes. P5 refuses to "fix" the ledger by inventing a shift the evidence does not clearly support.

Artifacts

P1–P5 pipeline (all live)