Towards human translations guided language discovery for ASR systems

Workshop on Spoken Language Technologies for Under-resourced Languages Pub Date : 1900-01-01 DOI:10.21437/Interspeech.2009-765

Sebastian Stüker, A. Waibel

{"title":"Towards human translations guided language discovery for ASR systems","authors":"Sebastian Stüker, A. Waibel","doi":"10.21437/Interspeech.2009-765","DOIUrl":null,"url":null,"abstract":"ABSTRACTNatural language processing systems, e.g for AutomaticSpeech Recognition (ASR) or Machine Translation (MT),havebeenstudiedonlyforafractionoftheapprox.7000lan-guagesthatexistintoday’sworld,themajorityofwhichhaveonlycomparativelyfewspeakersandfewresources.Thetra-ditionalapproachofcollectingandannotatingthenecessarytrainingdataisduetoeconomicconstraintsnotfeasibleformostofthem. AtthesametimeitisofvitalinteresttohaveNLPsystemsaddresspracticallyalllanguagesintheworld.New,efﬁcientwaysofgatheringtheneededtrainingmaterialhavetobefound.InthispaperweproposeanewtechniqueofcollectingsuchdatabyexploitingtheknowledgegainedfromHumansimultaneoustranslationsthathappenfrequentlyintherealworld. Toshowthefeasibilityofourapproachwepresentﬁrstexperimentstowardsconstructingapronuncia-tiondictionaryfromthedatagained.Index Terms — Automatic Speech Recognition, Lan-guage Discovery, Machine Translation, Under-ResourcedLanguages1. INTRODUCTION1.1. The Traditional Way to Acquire Training DataTraining large vocabulary continuous speech recognition(LVCSR) systems requires a number resources in the tar-getedlanguage. Fortrainingtheacousticmodelofarecog-nitionsystemlargeamountsoftranscribedaudiorecordingsofspeechareneeded.Thetrainingofthelanguagemodelre-quireslargeamountsofwrittentextinthetargetedlanguage.Whenusingphonemebasedacousticmodels,apronunciationdictionaryisneededthatmapsthewrittenrepresentationofawordtothesequenceofitsphonemeswhenbeingspoken.Approximately7,000languagesexisttoday,thecurrenteditionofEthnologue[1]lists7,299.Sofar,automaticspeechrecognition (ASR) systems and machine translation (MT)","PeriodicalId":190269,"journal":{"name":"Workshop on Spoken Language Technologies for Under-resourced Languages","volume":"31 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"23","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Workshop on Spoken Language Technologies for Under-resourced Languages","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.21437/Interspeech.2009-765","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}

引用次数: 23

Abstract

ABSTRACTNatural language processing systems, e.g for AutomaticSpeech Recognition (ASR) or Machine Translation (MT),havebeenstudiedonlyforafractionoftheapprox.7000lan-guagesthatexistintoday’sworld,themajorityofwhichhaveonlycomparativelyfewspeakersandfewresources.Thetra-ditionalapproachofcollectingandannotatingthenecessarytrainingdataisduetoeconomicconstraintsnotfeasibleformostofthem. AtthesametimeitisofvitalinteresttohaveNLPsystemsaddresspracticallyalllanguagesintheworld.New,efﬁcientwaysofgatheringtheneededtrainingmaterialhavetobefound.InthispaperweproposeanewtechniqueofcollectingsuchdatabyexploitingtheknowledgegainedfromHumansimultaneoustranslationsthathappenfrequentlyintherealworld. Toshowthefeasibilityofourapproachwepresentﬁrstexperimentstowardsconstructingapronuncia-tiondictionaryfromthedatagained.Index Terms — Automatic Speech Recognition, Lan-guage Discovery, Machine Translation, Under-ResourcedLanguages1. INTRODUCTION1.1. The Traditional Way to Acquire Training DataTraining large vocabulary continuous speech recognition(LVCSR) systems requires a number resources in the tar-getedlanguage. Fortrainingtheacousticmodelofarecog-nitionsystemlargeamountsoftranscribedaudiorecordingsofspeechareneeded.Thetrainingofthelanguagemodelre-quireslargeamountsofwrittentextinthetargetedlanguage.Whenusingphonemebasedacousticmodels,apronunciationdictionaryisneededthatmapsthewrittenrepresentationofawordtothesequenceofitsphonemeswhenbeingspoken.Approximately7,000languagesexisttoday,thecurrenteditionofEthnologue[1]lists7,299.Sofar,automaticspeechrecognition (ASR) systems and machine translation (MT)

查看原文本刊更多论文

面向人工翻译指导的ASR系统语言发现

摘要自然语言处理系统，如自动语音识别(ASR)或机器翻译(MT)，只针对当今世界上存在的大约7000种语言中的一小部分进行了研究，其中大多数语言只有相对较少的演讲者和较少的资源，传统的收集和注释必要的训练数据的方法由于经济限制对大多数语言来说是不可行的。AtthesametimeitisofvitalinteresttohaveNLPsystemsaddresspracticallyalllanguagesintheworld.New efficientwaysofgatheringtheneededtrainingmaterialhavetobefound.InthispaperweproposeanewtechniqueofcollectingsuchdatabyexploitingtheknowledgegainedfromHumansimultaneoustranslationsthathappenfrequentlyintherealworld。Toshowthefeasibilityofourapproachwepresentfirstexperimentstowardsconstructingapronuncia-tiondictionaryfromthedatagained。索引术语-自动语音识别，语言发现，机器翻译，资源不足语言。INTRODUCTION1.1。训练数据的传统获取方法训练大词汇量连续语音识别(LVCSR)系统需要大量的目标语言资源。Fortrainingtheacousticmodelofarecog-nitionsystemlargeamountsoftranscribedaudiorecordingsofspeechareneeded.Thetrainingofthelanguagemodelre-quireslargeamountsofwrittentextinthetargetedlanguage.Whenusingphonemebasedacousticmodels、apronunciationdictionaryisneededthatmapsthewrittenrepresentationofawordtothesequenceofitsphonemeswhenbeingspoken.Approximately7,000languagesexisttoday thecurrenteditionofEthnologue [1] lists7,299。迄今为止，自动语音识别(ASR)系统和机器翻译(MT)

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Workshop on Spoken Language Technologies for Under-resourced Languages

自引率

0.00%

发文量