That is actually an amazing idea. There is a new model that needs a voice sample of like 5 seconds and gives an almost perfect reproduction. Here is the link to the paper and a link to examples of what this does. I think this is the github repository of the implementation.
Mine is the name of one of the greatest elves of the first age in the Tolkien's Legendarium. The only one in the history that married a Maia, a god-like creature.