Hello,
as models evolve quickly. What are the current way(SOTA) to evaluate systems like speech recognition systems ? Are metrics like Word Error Rate or Character Error Rate still applicable to end-to-end system ? is there a link or paper that someone can provide.