Conference paper Closed Access

ESPnet How2 Speech Translation System for IWSLT 2019: Pre-training, Knowledge Distillation, and Going Deeper

Inaguma, Hirofumi; Kiyono, Shun; Soplin, Nelson Enrique Yalta; Suzuki, Jun; Duh, Kevin; Watanabe, Shinji


DataCite XML Export

<?xml version='1.0' encoding='utf-8'?>
<resource xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="http://datacite.org/schema/kernel-4" xsi:schemaLocation="http://datacite.org/schema/kernel-4 http://schema.datacite.org/meta/kernel-4.1/metadata.xsd">
  <identifier identifierType="DOI">10.5281/zenodo.3525560</identifier>
  <creators>
    <creator>
      <creatorName>Inaguma, Hirofumi</creatorName>
      <givenName>Hirofumi</givenName>
      <familyName>Inaguma</familyName>
      <affiliation>Kyoto University</affiliation>
    </creator>
    <creator>
      <creatorName>Kiyono, Shun</creatorName>
      <givenName>Shun</givenName>
      <familyName>Kiyono</familyName>
      <affiliation>RIKEN AIP &amp; Tohoku University</affiliation>
    </creator>
    <creator>
      <creatorName>Soplin, Nelson Enrique Yalta</creatorName>
      <givenName>Nelson Enrique Yalta</givenName>
      <familyName>Soplin</familyName>
      <affiliation>Waseda University</affiliation>
    </creator>
    <creator>
      <creatorName>Suzuki, Jun</creatorName>
      <givenName>Jun</givenName>
      <familyName>Suzuki</familyName>
      <affiliation>Tohoku University &amp; RIKEN AIP</affiliation>
    </creator>
    <creator>
      <creatorName>Duh, Kevin</creatorName>
      <givenName>Kevin</givenName>
      <familyName>Duh</familyName>
      <affiliation>Johns Hopkins University</affiliation>
    </creator>
    <creator>
      <creatorName>Watanabe, Shinji</creatorName>
      <givenName>Shinji</givenName>
      <familyName>Watanabe</familyName>
      <affiliation>Johns Hopkins University</affiliation>
    </creator>
  </creators>
  <titles>
    <title>ESPnet How2 Speech Translation System for IWSLT 2019: Pre-training, Knowledge Distillation, and Going Deeper</title>
  </titles>
  <publisher>Zenodo</publisher>
  <publicationYear>2019</publicationYear>
  <dates>
    <date dateType="Issued">2019-11-02</date>
  </dates>
  <language>en</language>
  <resourceType resourceTypeGeneral="Text">Conference paper</resourceType>
  <alternateIdentifiers>
    <alternateIdentifier alternateIdentifierType="url">https://zenodo.org/record/3525560</alternateIdentifier>
  </alternateIdentifiers>
  <relatedIdentifiers>
    <relatedIdentifier relatedIdentifierType="DOI" relationType="IsVersionOf">10.5281/zenodo.3525559</relatedIdentifier>
    <relatedIdentifier relatedIdentifierType="URL" relationType="IsPartOf">https://zenodo.org/communities/iwslt2019</relatedIdentifier>
  </relatedIdentifiers>
  <rightsList>
    <rights rightsURI="info:eu-repo/semantics/closedAccess">Closed Access</rights>
  </rightsList>
  <descriptions>
    <description descriptionType="Abstract">&lt;p&gt;This paper describes the ESPnet submissions to the How2 Speech Translation task at IWSLT2019. In this year, we mainly build our systems based on Transformer architectures in all tasks and focus on the end-to-end speech translation (E2E-ST). We first compare RNN-based models and Transformer, and then confirm Transformer models significantly and consistently outperform RNN models in all tasks and corpora. Next, we investigate pre-training of E2E-ST models with the ASR and MT tasks. On top of the pre-training, we further explore knowledge distillation from the NMT model and the deeper speech encoder, and confirm drastic improvements over the baseline model. All of our codes are publicly available in ESPnet.&lt;/p&gt;</description>
  </descriptions>
</resource>
257
22
views
downloads
All versions This version
Views 257257
Downloads 2222
Data volume 3.8 MB3.8 MB
Unique views 212212
Unique downloads 1919

Share

Cite as