Speech-to-text is amazing but misses so much emotional intent, very lossy I would love if there was a mode that was less lossy, even if it the generated text would look weird to a human and you'd never put it in a doc / slack message / etc