Improving vision-inspired keyword spotting using dynamic module skipping in streaming conformer encoder

Re-assign community