More ‘human-like’ speech

I find the current speech system to be too digitized and sometimes difficult to translate into effective feedback that I can learn from and implement any correction. It’s too digitized. Perhaps an integration with AI speech to make sound like a real human coach is talking?