Predictive model of intonation -- Audio samples