Predicting Severity of Mental State Using Vocal Output Characteristics