Acoustical and Environmental Robustness in Automatic Speech Recognition