Speech Recognition Over Digital Channels: Robustness and Sta