A novel bottleneck-BLSTM front-end for feature-level context modeling in conversational speech recognition.

Martin Wllmer, Bjrn W. Schuller, Gerhard Rigoll

Browse the full ASRU paper archive.