TortoiseTTS might be the closest https://github.com/neonbjb/tortoise-tts It's a few shot multi speaker model so you need just 3-4 little clips to train new voices.