As an owner, I want to look at my conversations and rename the speaker detected.
In the background, this will gather samples and add to speaker enrollment until it gets to high quality confidence and tell the user.
It will always take another sample though, as this means it didn't recongise the person