A time-weighted method for predicting the intelligibility of speech in the presence of interfering sounds