ysharma/gr-workflow-microduck-lab

Update README.md

voice: split transcript from status so a signed-out visitor is told why nothing happened

add voice input: record or upload a spoken routine, transcribed with Whisper

fix frame timing: sample rate is 50/every, so video played 0.8x and the telemetry time axis was 25% long

keep the skates on: coerce stray walking moves inside a skating routine

All 9 shipped policies: sit, roller skates + crouch-glide, head gestures; fix kick contact and near-plane clipping

render at 20 fps: ~20% off the round trip under osmesa

Microduck Lab: run the real shipped RL policies from a gr.Workflow

initial commit