ElevenLabs is probably the closest tool that can deliver exact human speech.
But it takes a lot more work than people care to realize.
Not only can you clone your voice but you can also play around with the advanced features regarding variability and similarity to get the AI to emote vocally.
You also have to play around with the script because ElevenLabs doesn't process all words strung in a certain order the same way.
With a lot of experimentation, you can get exact human speech.
This works great for YouTube videos as well as for AI-powered human presentation videos.
But it takes quite a bit of training work initially.
Once you get the hang of it, it's smooth and saves a tremendous amount of time and money down the road.