Text this: Throat and acoustic paired speech dataset for deep learning-based speech enhancement.