eduzhai > Applied Sciences > Engineering >

Online Automatic Speech Recognition with Listen Attend and Spell Model

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: The Listen, Attend and Spell (LAS) model and other attention-based automaticspeech recognition (ASR) models have known limitations when operated in a fullyonline mode. In this paper, we analyze the online operation of LAS models todemonstrate that these limitations stem from the handling of silence regionsand the reliability of online attention mechanism at the edge of input buffers.We propose a novel and simple technique that can achieve fully onlinerecognition while meeting accuracy and latency targets. For the Mandarindictation task, our proposed approach can achieve a character error rate inonline operation that is within 4 relative to an offline LAS model. Theproposed online LAS model operates at 12 lower latency relative to aconventional neural network hidden Markov model hybrid of comparable accuracy.We have validated the proposed method through a production scale deployment,which, to the best of our knowledge, is the first such deployment of a fullyonline LAS model.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...