Towards Effective and Compact Contextual Representation for Conformer Transducer Speech Recognition Systems

Open in new window