Conference paper Open Access

Robust Neural Machine Translation for Clean and Noisy Speech Transcripts

Di Gangi, Matti; Enyedi, Robert; Brusadin, Alessandra; Federico, Marcello

DataCite XML Export

<?xml version='1.0' encoding='utf-8'?>
<resource xmlns:xsi="" xmlns="" xsi:schemaLocation="">
  <identifier identifierType="DOI">10.5281/zenodo.3524947</identifier>
      <creatorName>Di Gangi, Matti</creatorName>
      <familyName>Di Gangi</familyName>
      <affiliation>Fondazione Bruno Kessler, Trento, Italy &amp; University of Trento, Italy</affiliation>
      <creatorName>Enyedi, Robert</creatorName>
      <affiliation>Amazon AI, East Palo Alto, USA</affiliation>
      <creatorName>Brusadin, Alessandra</creatorName>
      <affiliation>Amazon AI, East Palo Alto, USA</affiliation>
      <creatorName>Federico, Marcello</creatorName>
      <affiliation>Amazon AI, East Palo Alto, USA</affiliation>
    <title>Robust Neural Machine Translation for Clean and Noisy Speech Transcripts</title>
    <date dateType="Issued">2019-11-02</date>
  <resourceType resourceTypeGeneral="ConferencePaper"/>
    <alternateIdentifier alternateIdentifierType="url"></alternateIdentifier>
    <relatedIdentifier relatedIdentifierType="DOI" relationType="IsVersionOf">10.5281/zenodo.3524946</relatedIdentifier>
    <relatedIdentifier relatedIdentifierType="URL" relationType="IsPartOf"></relatedIdentifier>
    <rights rightsURI="">Creative Commons Attribution 4.0 International</rights>
    <rights rightsURI="info:eu-repo/semantics/openAccess">Open Access</rights>
    <description descriptionType="Abstract">&lt;p&gt;Neural machine translation models have shown to achieve high quality when trained and fed with well structured and punctuated input&amp;nbsp;texts. Unfortunately, the latter condition is not met in spoken language translation, where the input is generated by an automatic speech&amp;nbsp;recognition (ASR) system. In this paper, we study how to adapt a strong NMT system to make it robust to typical ASR errors. As in our application scenarios transcripts might be post-edited by human experts, we propose adaptation strategies to train a single system that can translate either clean or noisy input with no supervision on the input type. Our experimental results on a public speech translation data set show that adapting a model on a significant amount of parallel data including ASR transcripts is beneficial with test data of the same type, but produces a small degradation when translating clean text. Adapting on both clean and noisy variants of the same data leads to the best results on both input types.&lt;/p&gt;</description>
All versions This version
Views 689689
Downloads 161161
Data volume 22.5 MB22.5 MB
Unique views 659659
Unique downloads 143143


Cite as