Evelyn: Blink calibration did not pass
The separate Evelyn motion trial ran and failed complete-blink calibration: the two eyelid apertures reduced by about 1.7%, below the fixed 65% requirement. Jaw calibration was accepted, but the six integrated motion captures and video were not produced. Engineering audio-clock/neutral checks are software controls, not generated speech or physical sound verification. The completed saved-evidence review confirmed the failed blink criterion, recorded metrics and owned-process cleanup. No working blink or talking-person claim is justified.
Measured: Both eyelid apertures reduced by approximately 1.7%, below the predeclared 65% reduction requirement. The jaw calibration was accepted, but no integrated six-frame captures or completed video followed.
Not verified: working blink, synced speech, generated dialogue or animated talking head. The sound-clock tests are software timing checks, not proof of spoken audio.
Follow-up preparation (not a retest)
Subsequent October 20 task screenshots report that the initial source calibration explored only a 0.5° eyelid movement. A bounded correction now searches for sufficient closure, retaining stability and return-to-neutral tests. The code passed independent source review, but its fresh motion test had not started after RAM remained below the existing 8 GiB available-memory startup guard.
The separate three-angle head-render report documents actual successful static views. Static visibility and animated blinking are different acceptance gates.