Modelling Context in Automatic Speech Recognition by