On Using Transformers for Speech-Separation