<?xml version="1.0" encoding="utf-8"?>
<TEI xmlns="http://www.tei-c.org/ns/1.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:hal="http://hal.archives-ouvertes.fr/" xmlns:gml="http://www.opengis.net/gml/3.3/" xmlns:gmlce="http://www.opengis.net/gml/3.3/ce" version="1.1" xsi:schemaLocation="http://www.tei-c.org/ns/1.0 http://api.archives-ouvertes.fr/documents/aofr-sword.xsd">
  <teiHeader>
    <fileDesc>
      <titleStmt>
        <title>HAL TEI export of hal-03190119</title>
      </titleStmt>
      <publicationStmt>
        <distributor>CCSD</distributor>
        <availability status="restricted">
          <licence target="https://creativecommons.org/publicdomain/zero/1.0/">CC0 1.0 - Universal</licence>
        </availability>
        <date when="2026-05-16T07:02:08+02:00"/>
      </publicationStmt>
      <sourceDesc>
        <p part="N">HAL API Platform</p>
      </sourceDesc>
    </fileDesc>
  </teiHeader>
  <text>
    <body>
      <listBibl>
        <biblFull>
          <titleStmt>
            <title xml:lang="en">The Energy and Carbon Footprint of Training End-to-End Speech Recognizers</title>
            <author role="aut">
              <persName>
                <forename type="first">Titouan</forename>
                <surname>Parcollet</surname>
              </persName>
              <email type="md5">5f7f46eae91307e1e6c57795164be9ad</email>
              <email type="domain">gmail.com</email>
              <idno type="idhal" notation="string">titouan-parcollet</idno>
              <idno type="idhal" notation="numeric">174514</idno>
              <idno type="halauthorid" notation="string">44536-174514</idno>
              <idno type="ORCID">https://orcid.org/0000-0003-0672-1346</idno>
              <affiliation ref="#struct-100376"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Mirco</forename>
                <surname>Ravanelli</surname>
              </persName>
              <idno type="halauthorid">1569949-0</idno>
            </author>
            <editor role="depositor">
              <persName>
                <forename>Titouan</forename>
                <surname>Parcollet</surname>
              </persName>
              <email type="md5">5f7f46eae91307e1e6c57795164be9ad</email>
              <email type="domain">gmail.com</email>
            </editor>
          </titleStmt>
          <editionStmt>
            <edition n="v1" type="current">
              <date type="whenSubmitted">2021-04-06 09:12:51</date>
              <date type="whenModified">2021-04-08 09:41:49</date>
              <date type="whenReleased">2021-04-08 08:52:36</date>
              <date type="whenProduced">2021-04-06</date>
              <date type="whenEndEmbargoed">2021-04-06</date>
              <ref type="file" target="https://hal.science/hal-03190119v1/document">
                <date notBefore="2021-04-06"/>
              </ref>
              <ref type="file" subtype="author" n="1" target="https://hal.science/hal-03190119v1/file/INTERSPEECH___E2E_ASR_Carbon_Footprint-15.pdf" id="file-3190119-2799455">
                <date notBefore="2021-04-06"/>
              </ref>
            </edition>
            <respStmt>
              <resp>contributor</resp>
              <name key="729067">
                <persName>
                  <forename>Titouan</forename>
                  <surname>Parcollet</surname>
                </persName>
                <email type="md5">5f7f46eae91307e1e6c57795164be9ad</email>
                <email type="domain">gmail.com</email>
              </name>
            </respStmt>
          </editionStmt>
          <publicationStmt>
            <distributor>CCSD</distributor>
            <idno type="halId">hal-03190119</idno>
            <idno type="halUri">https://hal.science/hal-03190119</idno>
            <idno type="halBibtex">parcollet:hal-03190119</idno>
            <idno type="halRefHtml">2021</idno>
            <idno type="halRef">2021</idno>
            <availability status="restricted">
              <licence target="https://about.hal.science/hal-authorisation-v1/">HAL Authorization<ref corresp="#file-3190119-2799455"/></licence>
            </availability>
          </publicationStmt>
          <seriesStmt>
            <idno type="stamp" n="UNIV-AVIGNON">Université d'Avignon</idno>
            <idno type="stamp" n="LIA" corresp="UNIV-AVIGNON">Laboratoire Informatique d'Avignon</idno>
          </seriesStmt>
          <notesStmt/>
          <sourceDesc>
            <biblStruct>
              <analytic>
                <title xml:lang="en">The Energy and Carbon Footprint of Training End-to-End Speech Recognizers</title>
                <author role="aut">
                  <persName>
                    <forename type="first">Titouan</forename>
                    <surname>Parcollet</surname>
                  </persName>
                  <email type="md5">5f7f46eae91307e1e6c57795164be9ad</email>
                  <email type="domain">gmail.com</email>
                  <idno type="idhal" notation="string">titouan-parcollet</idno>
                  <idno type="idhal" notation="numeric">174514</idno>
                  <idno type="halauthorid" notation="string">44536-174514</idno>
                  <idno type="ORCID">https://orcid.org/0000-0003-0672-1346</idno>
                  <affiliation ref="#struct-100376"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Mirco</forename>
                    <surname>Ravanelli</surname>
                  </persName>
                  <idno type="halauthorid">1569949-0</idno>
                </author>
              </analytic>
              <monogr>
                <imprint/>
              </monogr>
            </biblStruct>
          </sourceDesc>
          <profileDesc>
            <langUsage>
              <language ident="en">English</language>
            </langUsage>
            <textClass>
              <classCode scheme="halDomain" n="info">Computer Science [cs]</classCode>
              <classCode scheme="halTypology" n="UNDEFINED">Preprints, Working Papers, ...</classCode>
              <classCode scheme="halOldTypology" n="UNDEFINED">Preprints, Working Papers, ...</classCode>
              <classCode scheme="halTreeTypology" n="UNDEFINED">Preprints, Working Papers, ...</classCode>
            </textClass>
            <abstract xml:lang="en">
              <p>Deep learning contributes to reaching higher levels of artificial intelligence. Due to its pervasive adoption, however, growing concerns on the environmental impact of this technology have been raised. In particular, the energy consumed at training and inference time by modern neural networks is far from being negligible and will increase even further due to the deployment of ever larger models. This work investigates for the first time the carbon cost of end-to-end automatic speech recognition (ASR). First, it quantifies the amount of CO2 emitted while training state-of-the-art (SOTA) ASR systems on a university-scale cluster. Then, it shows that a tiny performance improvement comes at an extremely high carbon cost. For instance, the conducted experiments reveal that a SOTA Transformer emits 50% of its total training released CO2 solely to achieve a final decrease of 0.3 of the word error rate. With this study, we hope to raise awareness on this crucial topic and we provide guidelines, insights, and estimates enabling researchers to better assess the environmental impact of training speech technologies.</p>
            </abstract>
          </profileDesc>
        </biblFull>
      </listBibl>
    </body>
    <back>
      <listOrg type="structures">
        <org type="laboratory" xml:id="struct-100376" status="VALID">
          <orgName>Laboratoire Informatique d'Avignon</orgName>
          <orgName type="acronym">LIA</orgName>
          <desc>
            <address>
              <addrLine>339 Chemin des Meinajaries Agroparc BP 1228 84911 Avignon cedex 9</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://lia.univ-avignon.fr/</ref>
          </desc>
          <listRelation>
            <relation active="#struct-195507" type="direct"/>
            <relation active="#struct-302221" type="direct"/>
          </listRelation>
        </org>
        <org type="institution" xml:id="struct-195507" status="VALID">
          <idno type="ROR">https://ror.org/00mfpxb84</idno>
          <orgName>Avignon Université</orgName>
          <orgName type="acronym">AU</orgName>
          <desc>
            <address>
              <addrLine>74 rue Louis Pasteur - 84 029 Avignon cedex 1</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.univ-avignon.fr/</ref>
          </desc>
        </org>
        <org type="institution" xml:id="struct-302221" status="VALID">
          <orgName>Centre d'Enseignement et de Recherche en Informatique - CERI</orgName>
          <desc>
            <address>
              <country key="FR"/>
            </address>
          </desc>
        </org>
      </listOrg>
    </back>
  </text>
</TEI>