New Approaches Towards Robust and Adaptive Speech Recognition