Currently, I’m using Deepspeech v4.1. I am a little confused about the word error rate. Is the reported word error rate based only on Deepspeech, or does any language model involved?
