Text this: Speech features analysis of the joint speech separation and automatic speech recognition model /