- location sensitive attention - attention masking - updated audio pre-processing (removed preemphasis and updated normalization) - testing with example senteces - optional eval-iteration