Maybe they are not training to make a TTS, but training on accents to make STT.