In the long run there is no way around it. To be honest if on device voices would be good enough then just sending the text version and let it read on the device in a somewhat natural voice would be kind of cool. I'm already using a special app which just reads any article I share with it in a fairly ok voice and I love this functionality, I don't need to sit down and read a long article on my phone, I can do something else and just make it read the article to me.

What I am worried about though are the people who do the research and who write the texts. Those sadly will also be replaced over time with machines and there is something we will really lose.