Predicting Individual Depression Symptoms from Acoustic Features During Speech