Text this: Robust parameterizations of Automatic Speech Recognition (ASR) in communications networks