1 min voice data can also be used to train a good TTS model! (few shot voice cloning). With inference command line tool included. - View it on GitHub