Facestar dataset. High quality audio-visual recordings of human conversational speech. - View it on GitHub