Scripts for running WER benchmarks are missing

[stri8ed]

On the releases page, there is a benchmarks table that includes a WER metric. I'm interested in reproducing those metrics, and running comparisons against different models that are not evaluated there. I looked at the benchmark scripts, but they only seem to measure the inference time. Can you share the scripts used for calculating the WER benchmarks? Thanks!