Modern Google-level speech-to-text models released

There is a related conversation here: https://news.ycombinator.com/item?id=24565831