Text this: Comparison of discrete transforms for deep‐neural‐networks‐based speech enhancement