Fetching the paper…

Transformer Transducer: A Streamable Speech Recognition Model with Transformer Encoders and RNN-T Loss · Around