Collection:
We present a beam decoding strategy for recurrent neural network transducers which has the characteristic that all competing hypotheses within the beam have the same alignment length (number of output symbols plus BLANK symbols). We contrast the proposed
- IEEE MemberUS $11.00
- Society MemberUS $0.00
- IEEE Student MemberUS $11.00
- Non-IEEE MemberUS $15.00
Videos in this product
Alignment-Length Synchronous Decoding For Rnn Transducer
We present a beam decoding strategy for recurrent neural network transducers which has the characteristic that all competing hypotheses within the beam have the same alignment length (number of output symbols plus BLANK symbols). We contrast the proposed