Keep our own books on the GPU budget

Report how fast the run actually was

Fewer GPU calls, faster attention, optional second pass

Mark the shakiest tenth, not the whole text

Estimate remaining time from measured pace

Cross-check, confidence, voice prints, quality, learning

Cut in silence, drop invented subtitle text

Make the glossary correct instead of decorate

Give Whisper its temperature fallback

Condense overlapping speech to meaningful passages

Split pure logic into kern.py; delete audio after analysis