N ov 2 01 9 DISTRIBUTED MICROPHONE SPEECH ENHANCEMENT BASED ON DEEP LEARNING