人人文库美如初恋！
资源预览需要最新版本的Flash Player支持。 您尚未安装或版本过低,建议您
中北大学2006届本科毕业设计说明书第1页共21页Theeffectofspeechrecognitiononworkingpostures,productivityandtheperceptionofuserfriendlinessElsbethM.deKortePietvanLingenAbstractAcomparative,experimentalstudywithrepeatedmeasureshasbeenconductedtoevaluatetheeffectoftheuseofspeechrecognitiononworkingpostures,productivityandtheperceptionofuserfriendliness.Fifteensubjectsperformedastandardisedtask,firstwithkeyboardandmouseand,afterasixweektrainingperiod,withspeechrecognition.Theuseofspeechrecognitionleadstoimprovedposturesofwrist,forearm,upperarmandshoulderandimprovementofneckmovementswhencomparedtotheuseofkeyboardandmouse.Althoughtheobservationmethodwasbasic,thisstudyprovidesinsightintothepotentialbenefitsspeechrecognitionhasforposture.However,productivitydecreasedformostsubjectsandspeechrecognitionappearstobeusableforspecifictasksonly.Fromtheperspectiveofproductivityandtheperceptionofuserfriendlinessfurtherdevelopmentofspeechrecognitionsoftwareisnecessary.Uptonow,speechrecognitionseemsespeciallybeneficialforpeoplewithWMSDcomplaints.1.IntroductionWorkrelatedmusculoskeletaldisordersWMSDarecommonamongVDUusersOfficeErgonomicsResearchCommittee,1998Gerretal.,2002BlatterandBongers,1999HealthCouncilofTheNetherlands,2000Ottenetal.,1998.VisualdisplayunitVDUusersareatriskofdevelopingneck,shoulder,wristandhandcomplaints.ThedurationofVDUwork,aswellasawkwardposturesofneck,shoulders,wristsandhandsareimportantriskfactors.Also,thelackofarmsupportmayleadtocomplaintsofneck,shoulder,armandhandPunnettandBerqvist,1997HalesandBernard,1996Marcusetal,2002BlatterandBongers,2002Ottenetal.,1998Tittiranondaetal.,1999OfficeErgonomicsResearchCommittee,1998.Withthearrivalofanewgenerationofcontinuousspeechrecognitionsoftware,speechrecognitionbecomesinterestingasanewkindofinputdevice.Itcanreplacebothmouseandkeyboard.Furthermore,becausespeechrecognitionsoftwarecanbeoperatedwithoutusingthehands,itenablesuserstomovefreelyattheworkplace.Avoidingawkwardposturesbecomesapossibility.Therefore,itisinterestingtoexplorewhetherspeechrecognitioncanreduceoneoftheWMSDriskfactors,thatisawkwardworkingpostures,andwhetheritmightplayaroleinprevention.However,researchonthistopicislimited.TherelationbetweenspeechrecognitionandboththeperceptionofuserfriendlinessandproductivityhasbeenstudiedpreviouslyBaberandNoyes,1996Baberetal.,1996Ponsioen,1999Bekkeretal.,1995NoyesandFrankish,1994.Thesestudieshaveshownthatthequalityofrecognitionofspeechisdependentonseveralfactors,forexamplethevoiceoftheuser,whichisinfluencedbyemotions,stress,coldorfatigue.Otherfactorsfoundtobeofimportanceare中北大学2006届本科毕业设计说明书第2页共21页backgroundnoiseandamountoftrainingtousespeechrecognition.Whenrecognitionqualitydecreases,thenumberoferrorsincreases,whichinfluencestheproductivityoftheusernegatively.Correctingerrorsintherightwayisveryimportantthesystemneedstobetrainedconstantlybytheusertoretainqualityandtoimprovethesystem.Becauseproductivityanduserfriendlinessdeterminetheactualpurchaseanduseofspeechrecognitionsoftwaretoalargeextent,itwasimportanttoinvolvethesefactorsinthisstudy.Thisstudywasdesignedtoevaluatetheeffectoftheuseofspeechrecognitiononworkingpostures,productivityanduserfriendlinessincomparisonwiththeuseofthetraditionalkeyboardandmouseduringVDUwork.2.Subjectsandmethods2.1ExperimentaldesignInacomparative,experimentalstudywithrepeatedmeasures,twokindsofinputdevicesweretested.Thetraditionalcombinationofkeyboardandmousewascomparedtospeechrecognitioninapreandposttest,respectively.Afterthepretest,thesubjectsreceivedtraininginusingspeechrecognitionbyacompanyspecialisedinspeechrecognitiontraining.During6weeksallsubjectsusedspeechrecognitionDragonNaturallySpeakingDutch3.6,whichcorrespondstotheEnglish3.6versionintheirdailywork.After6weeks,theposttesttookplace.Workingposturesandproductivityweremeasuredanduserfriendlinesswasassessed.Attheirownworkstation,thesubjectsperformedastandardisedcomputertaskwhichconsistedoftwosubtasks.ThefirstsubtaskconsistedofmakingandsendingemailsinMicrosoftOutlookfor5min,wordprocessingandtexteditinginMicrosoftWordfor9min,andchangingbetweenapplicationsMicrosoftExcel,PowerpointandInternetExplorerfor1min.Thetotaldurationofthissubtaskwas15min.Duringthissubtaskworkingpostureswererecordedonvideotape.Inthepretestthesubjectsusedkeyboardandmouse.Intheposttesttheywereallowedtousekeyboardandmousewhentheygotstuckitwasnotrealistictoaskthemtoperformthefirstsubtaskwithspeechrecognitiononly.Thesecondsubtaskconsistedofcopyingatextwithoutediting.Thistaskwasperformedforamaximumof10min.Thesubjectswereinstructedtoworkattheirownpace.Noinstructionsweregivenoncorrectingerrors.Thesecondsubtaskwasusedtomeasureproductivity.Consequently,thesubjectswererestrictedtospeechrecognitionintheposttest.2.2.SubjectsFifteensubjects,9menand6women,participatedinthestudy.TheyallworkedattheDutchMinistryofSocialAffairs.ThesubjectswereexperiencedVDUworkersandatleast50oftheirdailyworkconsistedofVDUwork.Theywereallnontouchtypists.EightsubjectssufferedfromWMSD.Sevensubjectswerefreeofinjury.Theiragevariedbetween25and55yearsMean42.5,SD8.8.Beforetheexperiment,participantsgaveinformedconsent.2.3.MeasuringmethodsWorkingpostureswererecordedonvideotapeforthetotaldurationofthefirstsubtask15min.ThecamerawasplacedatthedominantsideofthesubjectFigs.1.中北大学2006届本科毕业设计说明书第3页共21页WithTheObserverNoldustheimagesofthefirstconditionwereanalysedfordurationofposturesapercentageoftotaltimeandfrequencyofchangingposturesnumberofchangesperminute.TheobservedFig.1.Schematicdrawingofcameraplacement,rearview.Table1AnalysedposturesdominantsideofthebodyBodyregionVariablesNeckFlexion25°Rotationyes/noShoulderElevationliftingyes/noUpperarmFlexionsagittalplane0–20°20°ForearmPronation/supination/neutralArmsupportonworksurface/onarmrests/nosupportWristFlexion0–30°/flexion30°/extension0–30°/extension30°Radialdeviation/ulnardeviation/nodeviationpostureswereadoptedfromTheGuidelinesforPhysicalLoad,developedbyTNOWorkEmploymentvanderGrinten,1999.ThedependentvariablesareshowninTable1.DifferencesinworkingposturesweretestedwithaTtestforrepeatedmeasures.Significancelevelawassetat5twosided.Productivitymeasurementsweretakenfromthesecondsubtask.WithTheObserverNoldusthevideotapewasanalysedfornumberoferrorsperminuteandtimespentoncorrectingerrorspercentageoftotaltime.Furthermore,thetextwasanalysedforthenumberofwordsperminuteenteredcorrectly.DifferencesinproductivityweretestedwithaTtestforrepeatedmeasures.Finally,perceivedproductivitywasassessedbyaquestionHaveyoudonemore,lessorthesameamountofworksinceyoustartedusingspeechrecognitionAnswerstothisquestionarepresentedasfrequencies.Theperceptionofuserfriendlinessofspeechrecognitionwasassessedwithaquestionnaire.Thisquestionnairewasaimedatthe6weeksthesubjectswereusingspeechrecognitionintheirdailywork.Itwaspresentedaftertheposttest.Thequestionsinquiredabouthowdifficultitwastolearntousespeechrecognition,howlongittooktolearnitandpreferenceforinputdevice中北大学2006届本科毕业设计说明书第4页共21页keyboard,mouseorspeechrecognition.Furthermore,subjectswereaskedtonamecomputertaskstheyweresuitedtoperformwithspeechrecognitionandtaskstheywerenot.Finally,subjectswereaskedabouthowsatisfiedtheywerewithusingspeechrecognition.Datafromthequestionnairearepresentedasfrequencies.Twotimes,meetingswereorganisedtoexchangeexperiences.rmationfromthesemeetingsisalsoreported.3.Results3.1.WorkingposturesTheworkingpostureresultsFig.3a–hshow,withtheuseofspeechrecognitioncomparedtokeyboardandmouse,significantlylessneckflexion,lesspronationoftheforearmandlessulnardeviationofthewrist.Withtheuseofspeechrecognitionarmsupportwasusedmoreoftencomparedtotheuseofkeyboardandmouse.AscanbeseenfromFig.4,withtheuseofspeechrecognition,significantlylesschangeswerefoundinneckflexionpostures,neckrotationposturesandarmsupportcomparedtotheuseofkeyboardandmouse.Whenspeechrecognitionwasused,morechangeswerefoundinflexionposturesoftheupperarm.3.2.ProductivityThenumberofwordscorrectlyenteredperminutewassignificantlylargerwiththeuseofkeyboardandmousemean22.7,SD6.7comparedtotheuseofspeechrecognitionmean17.5,SD8.8.Althoughnosignificantdifferenceinthefrequencyoferrorswasfound,thetimespentoncorrectingerrorswassignificantlylongerwiththeuseofspeechrecognition.Withtheuseofkeyboardandmouse4.2ofthetimeisspentoncorrectingerrorsincontrastwith52.2withtheuseofspeechrecognition.Forproductivitymeasurements,alargevariationwasfoundbetweenthesubjects.Ninesubjectsreportedthattheywereasproductivewithusingspeechrecognitionastheywerewithusingkeyboardandmouse,5subjectswerelessproductiveandonesubjectwasmoreproductive.3.3.PerceptionofuserfriendlinessAscanbeseenfromTable2,learningtousespeechrecognitionsoftwarewas,,prettydifficult"accordingtosixsubjects,,,prettyeasy"accordingto8subjectsandveryeasy"accordingto1subject.Nobodyreportedthattheuseofspeechrecognitionsoftwarewas,,verydifficult".Theamountoftimethesubjectsspenttolearntousespeechrecognitionwasdescribedas,,prettymuch"by8subjectsand,,notsomuch"by7subjects.Nobodyspent,,verymuch"time.ThemajorityofthesubjectswithWMSDcomplaintswassatisfiedwiththeuseofspeechrecognition,putertasksthatwerereportedtobesuitabletobeperformedwithspeechrecognitionwereWordprocessing,makingnote"s,makinglongEmails,editingtext,makinglistsandchangingbetweenapplications.Tasksthatwerereportedtobeunsuitabletobeperformedwithspeech中北大学2006届本科毕业设计说明书第5页共21页recognitionwereworkinginspreadsheetssuchasExcel,Wordprocessinginaforeignlanguage,makingshortEmails,makingtablesandfigures,correctingtext,workingwithstatisticalapplicationssuchasSPSSanddealingwithacalendar.Fromthemeetingsitappearedthatspeechrecognitionwasappreciatedasacomplementaryinputdevicebecauseitprovidesameansforuserstovarytheirmethodofinteractingwiththecomputer.ForthesubjectswhosufferfromWMSDcomplaintsspeechrecognitionenablesthemtoremainworking.Oneimportantnegativeconsequenceoftheuseofspeechrecognitionwasthatfivesubjectsreportedtohavesorethroats.Fig.3a–h.DurationofposturesinthefirstsubtaskusingkeyboardandmouseKMorspeechrecognitionSRpo005po001.中北大学2006届本科毕业设计说明书第6页共21页Fig.4.Frequencyofposturechangesinthefirstsubtaskpo005.Table2Perceptionofuserfriendlinessofspeechrecognitionoutcomesrelatedtogender,ageandWMSDSubjectGenderAgeWMSDDifficultytolearnTimetolearnSatisfiedwithPreferedinputyearsVeryeasyVerymuchspeechdevicePrettyeasyPrettymuchrecognitionKeyboardPrettydifficultNotsomuchMouseVerydifficultSpeechRecognition1Female30NoPrettyeasyNotsomuchNoKeyboard2Male25NoPrettyeasyPrettymuchNoKeyboard3Male50NoVeryeasyNotsomuchYesSpeechrecognition4Male34NoPrettydifficultNotsomuchNoKeyboard5Male35NoPrettyeasyPrettymuchNoKeyboard6Male44NoPrettydifficultNotsomuchNoKeyboard7Male55NoPrettydifficultPrettymuchYesSpeechrecognition8Female44YesPrettyeasyPrettymuchYesSpeechrecognition9Female42YesPrettyeasyNotsomuchYesSpeechrecognition10Female49YesPrettydifficultPrettymuchYesSpeechrecognition11Female42YesPrettydifficultPrettymuchYesSpeechrecognition12Female54YesPrettyeasyPrettymuchYesMissing13Male40YesPrettyeasyPrettymuchNoKeyboard14Male41YesPrettydifficultNotsomuchYesMouse15Male52YesPrettyeasyNotsomuchNoKeyboard4.Discussion4.1.WorkingposturesTheuseofspeechrecognitionsoftwareleadstobetterposturesofwristandforearmbothareheldinneutralpositionmoreoften.Sincethewristismoreneutralwhenspeechrecognitionisused,itmaycontributetoareductionofWMSDriskfactors.Sluiteretal.2001,PunnettandBerqvist1997andHagbergetal.1995foundthatworkingwithwristsinanonneutralpostureisariskfactorfordevelopingWMSD,althoughMarcusetal.2002foundnoevidenceforarelationshipbetweenulnardeviationandWMSD.Werneretal.1997foundeffectsofpronation/supinationoncarpaltunnelpressure.TowhatextentthedecreaseinpronationoftheforearmcontributestoareductionofWMSDriskfactorsisnotknown.Also,upperarmandshoulderposturesareimprovedifspeechrecognitionisused,becausearmsupportsareusedmoreoften.Adecreaseinchangesbetweenthedifferentsupportedarmpostureswasfound.Itisnotclearifthatisdesirableornot,althoughincreasesinposturalfixityin中北大学2006届本科毕业设计说明书第7页共21页semistatictasksisgenerallynotviewedasapositivesituation.Inthisstudy,neckposturesimprovedbytheuseofspeechrecognitionsoftware.Furthermore,whenkeyboardandmousewereused,neckpostureswerechangedmoreoften8.2times/mincomparedtotheuseofspeechrecognitionsoftware2.6times/minLookingatthekeyboardlessoftenmaybeanexplanationforthesefindings,becauseitisnotasmuchanecessityfornontouchtypistswhenusingspeechrecognitionsoftware.Theincreaseinposturalchangesmayleadtolessstaticworkingpostures.SincestaticworkingposturesareariskfactorfordevelopingWMSDcomplaintsHealthCouncilofTheNetherlands,2000Sluiteretal.,2001PunnettandBerqvist,1997Hagbergetal.,1995,thesefindingsmaysuggestthat,fromtheperspectiveofpreventionofstaticneckpostures,itispreferabletousekeyboardandmouse,aslongasextremeendrangepositionsareavoided.Ontheotherhand,itcouldbearguedthatanincreaseinposturalchangesoftheneckleadstorepetitivemovements,alsoariskfactorfordevelopingWMSDcomplaintsHealthCouncilofTheNetherlands,2000Sluiteretal.,2001PunnettandBerqvist,1997Hagbergetal.,1995.Kilbom2000definesrepetitiousmovementsasmovementsthatoccurmorethan2times/min,althoughthisdoesnotrefertoneckmovements.Thelong–termeffectsofchangesinposturesoftheneckarenotfullyunderstood,yet.Allinall,whetheradecreaseinneckposturechangesfrom8.2to2.6times/minisanimprovementremainsunclear.Inthisstudyimprovedposturesofneck,upperarm,forearmandshoulderwerefound,althoughthenumberofsubjectswasrestrictedduetofinancialandorganizationallimitations.Thepowerofthisstudyisthereforerelativelylow.SignificanceleveltypeIoraerror,however,wassetattheconventionallevelof5twosided.Differencesandchangesinthisstudythataresignificant",willthereforeallhaveaprobabilityof5orlesstobetheresultofchanceorsamplingerror.Ifadifferenceorachangeisnotsignificant,however,thismaylargelybeduetolowpower.Thisstudyisnotwellequippedto,,prove"thataparticularchangefrompretoposttestdoesnotexistinreality.Furthermore,theobservationswereimpreciseinsomecasesduetotheuseofonlyonecameraandlargeangularintervals.Also,therewerelargevariationsintheextenttowhichthesubjectsusedbothkeyboardandmousealongwithspeechrecognitioninthefirstsubtask,inwhichtheywereaskedtousespeechrecognitionasmuchaspossible.Allthesame,theresultsprovideusefulinsightintothepotentialbenefitsspeechrecognitionhasforposture.4.2.ProductivityInWordprocessing,themeannumberofwordsenteredcorrectlyperminuteislowerwiththeuseofspeechrecognitioncomparedtotheuseofkeyboardandmouse.However,whentheywereusingspeechrecognition,thesubjectsspenthalfthetimeoncorrectingerrors,comparedtoonly4ofthetimewhenusingkeyboardandmouse.Thisindicatesthattheremaybeleftmuchroomforimprovementsinproductivityofspeechrecognition.Productivitywouldimprovesubstantially,ifthenumberoferrorsandcorrectingtimedecreased.Thefactthatsomeofthesubjectsreachedahighernumberofwordsenteredcorrectlyperminutesupportstheideathatthereisroomforimprovement.Thepreandposttestwerenotpresentedatthesamemomentoftheday.Thus,therewasnotaccountedforfatigueeffectscausedbyworkactivities.TheresultsofthisstudyarecomparabletothoseofBekkeretal.1995.Theycomparedmouseusewiththeuseofspeechrecognitionandfoundhighererrorrateswhenspeechrecognitionwasused.Nevertheless,speechrecognitionhasdevelopedconsiderablyinqualityofrecognitionsince中北大学2006届本科毕业设计说明书第8页共21页thestudyofBekkeretal.1995wasperformed.Theresultsonproductivitymighthavebeeninfluencedbytheversionofthespeechrecognitionsoftware.ThesubjectswerenotabletousethemostrecentversionofthespeechrecognitionsoftwarebecauseitdidnotmatchintheotherofficeapplicationsusedattheDutchMinistryofSocialAffairs.Ifthelatestversionofthespeechrecognitionsoftwarewasused,theproductivityresultsmighthavebeenbetter.Also,thenumberofhoursthesubjectsspentontrainingwithspeechrecognitionsoftwaremayhaveinfluencedtheproductivityresults.Almostnoneofthesubjectscompletedtheplanned40hoftraining,oftenasaresultofdeadlinesandrushjobs.Also,largedifferenceswerefoundinthenumberofhoursspentlearningspeechrecognition.Thiscouldhaveledtolargedifferencesinskill.Ifthesubjectshadcompletedthe40hoftraining,theproductivityresultsmighthavebeenbetter.Theplanned40hoftrainingwithspeechrecognitionsoftwareduringdailyworkappearedtobeagreatdemandontheproductivityoftheorganisationasawhole.Nonetheless,forreintegratingemployeeswithWMSDcomplaints,speechrecognitioncouldbetheonlyoptionforbeingabletoworkatallwithacomputer.4.3.PerceptionofuserfriendlinessAfairlylargepartofthesubjects6subjectsreportedhatlearningtousespeechrecognitionsoftwarewasprettydifficult".Inaddition,ittookquitesometimetogetspeechrecognitionundercontrol.Nevertheless,6subjectspreferredspeechrecognitiontokeyboardormouse,themajorityofthesesubjectssufferedfromWMSDcomplaints.Unfortunately,therewerenoquestionsonthepreferredcombinationsofinputdevices.Sincespeechrecognitionappearstobeusableorspecifictasksonlyandisusablealongwithotherinputdevices,suchaskeyboardandmouse,itsvaluemightbeagoodcomplementtoexistinginputdevices.Fivesubjectsreportedasorethroatafterusingspeechrecognition.Furtherresearchisnecessarytodeterminetheeffectofspeechrecognitiononcomplaintswithvoiceorthroat.Afewarticlesgointothevoicedemandsassociatedwithextendedorfrequentuseofspeechrecognition,whichcanbehighandmayplaceusersatriskforvocaldifficulties,likemuscletensiondysphonia.Kambeyandaetal.,1997Olsonetal.,2004Haxeretal.,2001Williams,2003JuulKristensenetal.,2004.ExceptthearticleofJuulKristensenetal.2004whichisacomparativeexperimentalstudy,theseareallcasereports.Anoteworthyfindingarethechangesinspeechpatternswiththeuseofspeechrecognition.Forexample,Olsonetal.2004reportedthatallcaseshadnormalvoicewhenusingeverydayspeech,butspeakingintothecomputerresultedintherapidonsetofaperiodicity,strain,andadecreaseinfundamentalfrequency.ChangesinspeechpatternswerealsofoundinEMGmeasurementsofvoicerelatedmuscleswiththeuseofspeechrecognitionJuulKristensenetal.,2004.However,furtherstudiesareneededtoinvestigatethefindingsinthesearticles.Inthemeantime,itisrecommendedthatusersbecomeinformedabouttheunnaturalspeechpatternsusedwithspeechrecognition,learntousegoodvocalhygiene,suchasperformingwarmupandcooldownvoiceexercisesandusealternatemethodsofinputalongwiththespeechrecognitionproductKambeyandaetal.,1997.Althoughthenumberofsubjectshasbeensmall,interestingdatawasfoundthatmayleadtofurther,moreextensiveresearch.Fromtheoutcomesonuserfriendliness,forexample,itappears中北大学2006届本科毕业设计说明书第9页共21页thatspeechrecognitionisratedbetterbysubjectswithWMSDcomplaints.However,thenumberofsubjectswastoosmalltomakegroupcomparisons.ItmaybeinterestingforfutureresearchtocomparesubjectswithandsubjectswithoutWMSDcomplaintsastheyusespeechrecognitionsoftware.5.ConclusionSpeechrecognitionsoftwareappearstoimproveworkingposturesduringVDUworkandtherefore,itmayplayaroleinpreventionofWMSD.However,itseemsespeciallybeneficialforpeoplewithWMSDcomplaints,whocouldcontinueworkingwiththeuseofspeechrecognition.Furtherresearchisneededtostudyriskfactorsforvocaldifficultieswiththeuseofspeechrecognition.Whilecurrentspeechrecognitionisnotrecommendedforgeneraluse,itmaybeausefulcomplementtootherinputdevicesforpeoplewithoutWMSDcomplaints.Fromtheperspectiveofproductivityandtheperceptionofuserfriendlinessfurtherdevelopmentofspeechrecognitionsoftwareisnecessary.ReferencesBaber,C.,Noyes,J.,1996.Automaticspeechrecognitioninadverseenvironments.HumanFactors381,142–155.Baber,C.,Mellor,B.,Graham,R.,Noyes,J.M.,Tunley,C.,1996.Workloadandtheuseofautomaticspeechrecognitiontheeffectoftimeandresourcedemands.SpeechCommun.20,37–53.Bekker,M.M.,VanNes,F.L.,Juola,J.F.,1995.Acomparisonofmouseandspeechinputcontrolofatextannotationsystem.BehaviourInform.Technol.141,14–22.Blatter,BM.,Bongers,P.M.,1999.WorkrelatedneckandupperlimbsymptomsRSIhighriskoccupationsandriskfactorsintheDutchworkingpopulation.Publ.no.r9800293.TNOWorkandEmployment,Hoofddorp.Blatter,B.M.,Bongers,P.M.,2002.Durationofcomputeruseandmouseuseinrelationtomusculoskeletaldisordersofneckorupperlimb.Int.J.Ind.Ergon.30,295–306.Gerr,F.,Marcus,M.,Ensor,C.,Kleinbaum,D.,Cohen,S.,Edwards,A.,Gentry,E.,Ortiz,D.J.,Monteilh,C.A.,2002.ProspectivestudyofcomputerusersI.Studydesignandincidenceofmusculoskeletalsymptomsanddisorders.Am.J.Ind.Med.414,221–235.VanderGrinten,M.P.,1999.Arbouwguidelinesforphysicalloadwhenusinghandtoolsandhandlinginpackaginginthebuildingindustry.TNOWorkandEmployment,Hoofddorp,TheNetherlandsInDutchPubl.no.R0172.Confidential.Hagberg,M.,Silverstein,B.,Wells,R.,Smith,M.J.,Hendrick,H.W.,Carayon,P.,Perusse,M.,1995.WorkrelatedmusculoskeletaldisordersWMSDsareferencebookforprevention.TaylorFrancisLtd.,London.Hales,T.R.,Bernard,B.P.,1996.Epidemiologyofworkrelatedmusculoskeletaldisorders.Orthop.Clin.N.Am.274,679–709.Haxer,M.J.,Guinn,L.W.,Hogikyan,N.D.,2001.Useofspeechrecognitionsoftwareavocal中北大学2006届本科毕业设计说明书第10页共21页endurancetestforthenewmillenniumJ.Voice152,231–236.HealthCouncilofTheNetherlandsRSI,2000.TheHagueHealthCouncilofTheNetherlands,publicationno.2000/22InDutch.JuulKristensen,B.,Laursen,B.,Pilegaard,M.,Jensen,B.R.,2004.Physicalworkloadduringuseofspeechrecognitionandtraditionalinputdevices.Ergonomics472,119–133.Kambeyanda,D.,Singer,L.,Cronk,S.,1997.Potentialproblemsassociatedwithuseofspeechrecognitionproducts.Assist.Technol.92,95–101.Kilbom,A.,2000.RepetitiveworkoftheupperextremityPartIguidelinesforthepractionerandPartIIthescienticbasisfortheguide.InMital,A.,Kilbom,A.,Kumar,S.Eds.,ErgonomicsGuidelinesandProblemSolving.Elsvevier,Amsterdam.Marcus,M.,Gerr,F.,Monteilh,C.,Ortiz,D.J.,Gentry,E.,Cohen,S.,Edwards,A.,Ensor,C.,Kleinbaum,D.,2002.AprospectivestudyofcomputerusersIIposturalriskfactorsformusculoskeletalsymptomsanddisorders.Am.J.Ind.Med.41,236–249.Noyes,J.M.,Frankish,C.R.,1994.Errorsanderrorcorrectioninautomaticspeechrecognitionsystems.Ergonomics–1957.OfficeErgonomicsResearchCommittee,1998.Musculoskeletaldisordersintheofficeworkplace.FindingsoftheOfficeErgonomicsResearchCommitteeOERC.YarmouthPort,MA.Olson,D.E.,Cruz,R.M.,Izdebski,K.,Baldwin,T.,2004.Muscletensiondysphoniainpatientswhousecomputerizedspeechrecognitionsystems.EarNoseThroatJ.833,195–198.Otten,F.,Bongers,P.,Houtman,I.,1998.TheriskofdevelopingRSIinTheNetherlands,dataformthecontinuousqualityoflifesurvey,1997.Maandberichtgezondheidsstatistiek17,5–19InDutch.Ponsioen,I.H.J.,1999.Automaticspeechrecognitionstateoftheartandexploringthefuture.TijdschriftvoorErgonomie,140–147InDutch.Punnett,L.,Berqvist,U.,1997.VisualDisplayUnitWorkandupperextremitymusculoskeletaldisorders.Areviewofepidemiologicalfindings.NationalInstituteforWorkingLife1–160.Sluiter,J.K.,Rest,K.M.,FringsDresen,M.H.,2001.Criteriadocumentforevaluatingtheworkrelatednessofupperextremitymusculoskeletaldisorders.Scand.J.WorkEnviron.Health.27Suppl1,1–102.Tittiranonda,P.,Burastero,S.,Rempel,D.,1999.Riskfactorsformusculoskeletaldisordersamongcomputerusers.Occupat.Med.141,17–38.Werner,R.,Armstrong,T.,Bir,C.,Aylard,M.,1997.Intracarpalcanalpressurestheroleofnger,hand,wristandforearmposition.Clin.Biomech.12,44–51.Williams,N.R.,2003.VoicerecognitionproductsanoccupationalriskforuserswithULDs中北大学2006届本科毕业设计说明书第11页共21页Occupat.Med.537,452–455.中北大学2006届本科毕业设计说明书第12页共21页语音识别在工作姿势、提高生产率和用户友好界面的作用摘要在评价语音识别对工作姿势、提高生产率和用户友好界面的作用上做了相当多的实验性的研究以及重复性测试。首先在对键盘和鼠标以及语音识别进行六个星期训练期以后,实现了十五个主题执行了一项规范化的任务。对语音识别的用途导致手腕的改善、姿势、前臂、膀臂和脖子运动的肩膀和改善当与对键盘和鼠标的用途比较。虽然观察方法是基本的,这项研究提供洞察入潜力好处语音识别有为姿势的。但是,生产力被减少为多数主题和语音识别看来是能用的为具体任务唯一。从生产力透视和语音识别的用户友好界面看更加进一步的发展悟性软件是必要的。到现在,语音识别似乎特别有利于人们对WMSD的抱怨。1介绍与工作相关的musculoskeletalWMSD在VDU用户办公室人体工程的研究Committee1998年之中Gerr等2002年Blatter和Bongers1999年荷兰的健康委员会,2000年Otten等,1998中很普遍。显示器单位VDU用户是在危险中开发脖子、肩膀、腕子和手怨言。VDU工作的期间,并且脖子笨拙姿势,肩膀、腕子和手是重要风险因素。并且,缺乏胳膊支持也许导致脖子、肩膀、胳膊和手怨言Punnett和Berqvist1997年Hales和Bernard1996年Marcus等2002年Blatter和Bongers2002年Ottenetal.1998年Tittiranonda等1999年办公室人体工程的研究委员会,1998。以连续的新一代语音识别软件的到来,使语音识别作为一新输入装置变得有趣。它可能替换鼠标和键盘。此外,因为语音识别软件可能实现无手管理,它使用户自由地搬走在工作场所。避免笨拙姿势成为可能。所以,它是有趣探索是否语音识别可能减少WMSD风险因素的当中一个,是笨拙工作姿势,并且是否它也许充当在预防的一个角色。但是,对这个题目的研究是有限的。语音识别和用户友好界面及生产力之间的联系早已研究过Baber和Noyes1996年Baber等1996年Ponsioen1999年Bekker等1995年Noyes和法兰克人,1994。这些研究表示,讲话的认识的质量依靠几个因素,例如声音用户,中北大学2006届本科毕业设计说明书第13页共21页被情感、重音、寒冷或疲劳影响。其它被发现的重要因素是背景噪声和相当数量训练使用语音识别。当认识质量减少,错误的数量增加,消极地影响用户生产力。改正错误用正确的方式非常重要系统需要由用户经常训练保留质量和改进系统。由于生产力和用户友好界面确实在对语音识别软件的实际购买应用到大规模范围,在这项研究中介入这些因素就显得重要了。这项研究被设计评估在VDU工作期间对语音识别在工作姿势、生产力和用户友好界面的用途与对传统键盘和鼠标的用途比较。2主题和方法2.1实验性设计在相当多的实验性研究及重复的测试对两项输入装置进行测试。键盘和鼠标的传统组合与语音识别比较各自在前和进一步测验。在预告测验以后,主题接受了训练在使用语音识别由公司被专门研究语音识别训练。在6个星期期间他们的每日工作所有主题使用了语音识别龙自然地讲荷兰语3.6,对应于英语3.6版本。在6个星期以后,进一步测验进行。工作姿势和生产力被测量了并且用户友好度得到了评定估计。在他们自己的工作站,主题执行了包括二个子任务的一项规范化的计算机任务。第一子任务包括的制造和发送在MicrosoftOutlook为5min的电子邮件,在MicrosoftWord为9min的文字处理和正文编辑,和改变在申请之间微软擅长,力量点和InternetExplorer对1min。这个子任务的总期间是15分钟。在这子任务期间工作姿势被记录了在录影带。在预告测验主题使用了键盘和鼠标。在进一步测验他们被允许使用键盘和鼠标要求他们仅仅以语音识别执行第一子任务是不现实的。第二个子任务包括复制没有编辑的文本。这项任务执行的最大值为10min。主题被指示运作在他们自己的节奏。没有指示显示在改正错误上。第二个子任务使用测量生产力。结果,主题被限于语音识别的进一步测验中。2.2主题9名男子和6名妇女对十五个主题参加了研究。他们全都在荷兰部运作社会中北大学2006届本科毕业设计说明书第14页共21页事务。主题是老练的VDU工作者并且他们的每日工作至少50包括了VDU工作。他们是所有非接触打字员。八个主题遭受了WMSD。七个主题免于伤害。他们的年龄变化在25和55年手段42.5,SD8.8之间。在实验之前,参加者允许被通告。2.3测量的方法在总期间15分钟的第一子任务里工作姿势被记录在录影带。照相机被安置了在主题的显要一边Figs.1。图1.照相机安置,背面图概要图画与观察员Noldus对于姿势的第一个情况的图象分析总时间的百分比的改变的姿势变动的数字期间且频率每分钟。姿势的观察被采取了物理装载的指南,由TNO工作就业vanderGrinten1999开发。因变量被显示在表1。以T测试对工作姿势上的区别作重复的测试。意义水平a被设置了在5双面。Table1AnalysedposturesdominantsideofthebodyBodyregionVariablesNeckFlexion25°Rotationyes/noShoulderElevationliftingyes/noUpperarmFlexionsagittalplane0–20°20°ForearmPronation/supination/neutralArmsupportonworksurface/onarmrests/no中北大学2006届本科毕业设计说明书第15页共21页supportWristFlexion0–30°/flexion30°/extension0–30°/extension30°Radialdeviation/ulnardeviation/nodeviation生产力测量被采取了第二个子任务。与观察员Noldus录影带分析对于错误的数字每分钟和时间在改正上花费错误总时间的百分比。此外,每分钟输入正确的数量的词对文本进行分析。在生产力上的区别了以T测试为重覆的措施做测试。最后,被察觉的生产力产生出问题您完成了更多,或同样相当数量工作自从您开始使用语音识别对这个问题的答复被提出作为频率。语音识别的用户友好的悟性被估计了与查询表。这张查询表目标在于语音识别在他们的每日工作的6个星期中主题的使用。它被提出了在进一步测验以后。问题询问了关于它将学会使用语音识别的困难,多长时间需要学会它和特选为输入装置键盘、鼠标或语音识别。此外,他们适合执行语音识别的任务的主题不是请求命名计算机任务。终于,主题被询问他们是否对使用语音识别感到满意。数据从查询表被提出作为频率。组织了两次会议交换经验。首次会议被组织在训练期的第三个星期为了与语音识别软件分享第一经验。第二次会议被组织了在实验发生为了评估对语音识别软件的用途和对交换意见之后。信息来自这些会议并且被报告。3.结论3.1工作姿势工作姿势结果图3.ah显示,通过语音识别的用途与键盘和鼠标的对比,极大减少了脖子的弯曲、前臂的内旋和手腕骨的背离。以对语音识别胳膊的用途支持与键盘和鼠标的用途做了更多的比较。从图4能看出,在语音识别的应用中,在脖子弯曲姿势、脖子自转姿势和胳膊支持与对键盘和鼠标的用途比较中没有发现较大的变动。当语音识别被使用了,更多变化被发现了在膀臂的弯曲姿势上。3.2生产力词的数量正确地输入每分钟是显着大的以对键盘和鼠标手段22.7,SD6.7中北大学2006届本科毕业设计说明书第16页共21页的用途与对语音识别比较手段17.5,SD8.8的用途。虽然在错误频率上的重大区别未被发现,但花费在修正错误上的时间和语音识别的使用一样长。以对键盘和鼠标4.2的用途时间与花费在错修正52.2对比的语音识别的用途。为生产力测量,在主题之间发现了大的变异,他们是一样有生产力的以使用语音识别象他们以使用键盘和鼠标,5个主题是没有生产力的并且一个主题是更加有生产力的。3.3用户友好的悟性从表2能看出,学会使用语音识别软件是相当困难根据六个主题,相当容易根据8个主题和非常容易根据1个主题。没人报告,对语音识别软件的用途是非常困难。时间主题花费在学会使用语音识别上几乎被描述了很非常作为8个主题和不非常作为7主题。没人花费了非常多时间。主题的多数以WMSD怨言满足了对语音识别的用途,与伤害主题对比。主题以WMSD怨言比任意是伤害的主题经常并且更喜欢语音识别当输入装置。被报告是适合用语音识别执行的计算机任务是字词处理,做笔记的,做长的电子邮件,编辑文本,做名单和改变在应用之间。被报告不适合以语音识别的执行任务运作在报表里譬如擅长,字词处理在一种外语,做短的电子邮件,做桌和图,改正文本,运作以统计应用譬如SPSS和处理日历。从会议看起来,语音识别被赞赏了作为一种补全输入装置因为它为用户提供可变化的方法使他们与计算机相处融洽。为遭受WMSD怨言语音识别的主题使他们依然是工作。对语音识别的用途的一重要消极后果是,五个主题报告有疼痛喉头。中北大学2006届本科毕业设计说明书第17页共21页图3ah姿势的期间在第一子任务使用键盘和鼠标KM或语音识别SRpo005po001图4姿势变化频率在第一子任务上po005语音识别结果的用户友好的悟性与性别、年龄和WMSD关系SubjectGenderAgeWMSDDifficultytolearnTimetolearnSatisfiedwithPreferedinputyearsVeryeasyVerymuchspeechdevicePrettyeasyPrettymuchrecognitionKeyboardPrettydifficultNotsomuchMouseVerydifficultSpeechRecognition1Female30NoPrettyeasyNotsomuchNoKeyboard2Male25NoPrettyeasyPrettymuchNoKeyboard3Male50NoVeryeasyNotsomuchYesSpeechrecognition中北大学2006届本科毕业设计说明书第18页共21页4Male34NoPrettydifficultNotsomuchNoKeyboard5Male35NoPrettyeasyPrettymuchNoKeyboard6Male44NoPrettydifficultNotsomuchNoKeyboard7Male55NoPrettydifficultPrettymuchYesSpeechrecognition8Female44YesPrettyeasyPrettymuchYesSpeechrecognition9Female42YesPrettyeasyNotsomuchYesSpeechrecognition10Female49YesPrettydifficultPrettymuchYesSpeechrecognition11Female42YesPrettydifficultPrettymuchYesSpeechrecognition12Female54YesPrettyeasyPrettymuchYesMissing13Male40YesPrettyeasyPrettymuchNoKeyboard14Male41YesPrettydifficultNotsomuchYesMouse15Male52YesPrettyeasyNotsomuchNoKeyboard4讨论4.1工作姿势对语音识别软件的用途引导改善手腕和前臂姿势两个部位经常在中间位置运动。因为手腕是更加中立的当语音识别被使用,它也许对WMSD风险因素减少贡献。Sluiter等2001,Punnett和Berqvist1997并且Hagberg等1995发现工作与腕子在一个非中立姿势是风险因素为开发WMSD,虽然Marcusetal.2002没有发现证据为一个关系在尺骨的偏差和WMSD之间。Werner和al.1997内旋旋后的被发现的作用在腕骨隧道压力。在何种程度上在前臂的内旋的减退对WMSD风险因素减少贡献不为人所知。并且,语音识别被使用改进了膀臂和肩膀姿势,因为胳膊支持经常被使用。在变动的减退在不同的援军姿势之间被发现了。它不是确切如果那是中意的或不是,虽然在姿势固定性的增量在半静态任务下一般不被观看作为一个正面情况。在这项研究中被改进摆姿势脖子,膀臂,前臂和肩膀被发现了,虽然主题的数量有限归结于财政和组织局限。这项研究的力量是因此相对地降低。意义水平型I或错误,然而,被设置了在常规水平的5双面。在学习上的区别和变化中北大学2006届本科毕业设计说明书第19页共21页是重要的,因此所有将有5或者可能性是机会或抽样误差的结果。如果区别或变动不是重大的,然而,这也许主要归结于低力量。这项研究不是对装备好的证明,实际上特殊变动从前对进一步测验不存在。此外,观察不精确在某些情况下归结于对仅仅一台照相机和大有角间隔时间的用途。并且,在主题在第一子任务使用键盘和鼠标语音识别一起,他们请求使用语音识别尽量的程度上大变化。同样,结果提供有用的潜在的好处语音识别有姿势的。4.2生产力在文字处理,词的数量正确地被输入每分钟是低以对语音识别的用途与对键盘和鼠标的用途比较。但是,当他们使用语音识别,主题在改正上花费了一半时间错误,而对此只有4键盘和鼠标。这表明,那里也许为改善在语音识别生产力留有更多的空间。如果错误和改正时间的数字减少了,生产力极大地会改善。事实一些主题到达了词的一个更高的数字正确地被输入每分钟支持想法,有空间提供改善。前和进一步测验未在同一天出现。因而,那里未占疲劳作用由工作活动造成。这项研究的结果与那些是可比较的Bekker等1995。他们对鼠标用途与对语音识别的用途比较和发现了更高的误差率当语音识别被使用了。然而,自从Bekker等1995的对语音识别进行了研究后,可观地显现了在认识的质量上。结果在生产力也许被语音识别软件的版本已经影响了。主题没有能使用语音识别软件的较新版本因为它没有匹配在其它办公室应用被使用在荷兰部社会事务。如果语音识别软件的最新的版本被使用了,生产力结果也许已经是更好。并且,几小时的数量主题花在训练以语音识别软件上也许影响了生产力结果。几乎无主题完成了训练计划的40h,经常由于最后期限和仓促工作。并且,大区别被发现了在几小时被花费的学习的语音识别的数字。这能导致了在技巧上的大区别。如果主题完成了训练40h,生产力结果也许已经会更好。计划的40h训练以语音识别软件在每日工作期间看来在组织的生产力整体上是巨大的需求。仍然,雇员对WMSD的怨言,语音识别是能为计算机工作的唯一的选择。4.3用户友好的悟性主题6个主题相当大部份报告了帽子学会使用语音识别软件非常困难。中北大学2006届本科毕业设计说明书第20页共21页另外,需要了相当某个时候得到语音识别在控制之下。然而,6个主题更喜欢语音识别操作或鼠标,多数这些主题遭受了WMSD怨言。不幸地,没有对输入装置的更喜欢的组合的问题。因为语音识别看来是能用或具体任务唯一和是能用与其它输入装置一起,譬如键盘和鼠标,它的价值也许是对现有的输入装置最好的补全。五个主题以后报告了使用语音识别后疼痛喉头。进一步研究是必要确定语音识别在怨言与声音或喉头的作用。几篇文章进入声音要求与相关对语音识别的延长或频繁用途,是高的,也许安置用户在危险中为声音困难,象肌肉紧张、发音困难。Kambeyanda等1997年Olson等2004年Haxer等2001年威廉斯2003年JuulKristensen等,2004。除了是一项比较实验性研究的文章JuulKristensen等2004,这些是所有案件报告。一显著发现是变化在讲话样式上以对语音识别的用途。例如,Olson等2004报告,所有案件当使用每天讲话都有正常声音,但讲话入计算机导致一个周期性、张力和在基频的减退的迅速起始。变化在讲话样式上并且被发现了在声音相关的肌肉的EMG测量以对语音识别JuulKristensen等2004的用途。但是,进一步研究是需要的调查研究结果在这些文章里。同时,它建议,用户变得消息灵通关于不自然的讲话样式被使用于语音识别,学会使用好声音卫生学,譬如执行准备和凉快下来的声音锻炼和与语音识别产品Kambeyanda等1997一起使用输入交错法。虽然主题的数量小,也许导致进一步的有趣的数据被发现了且用于广泛的研究。从结果在用户友好界面看起来,语音识别是额定的更好由主题对WMSD的抱怨。但是,主题的数量太小以至于不能做小组比较。当他们使用语音识别软件时,未来研究主题与没有WMSD抱怨主题的比较是十分有趣的。5结论语音识别软件看上去在VDU工作期间改进工作姿势因此它也许充当在WMSD的预防的一个角色。但是,它似乎特别有利于那些能持续使用语音识别的人对WMSD的抱怨。进一步研究需要学习的是语音识别使用中的风险因素为声音制造的困难。而当前的语音识别不推荐为普遍的使用,这也许是对其它输入中北大学2006届本科毕业设计说明书第21页共21页装置人们没有WMSD怨言来说只是个有用的补充,。从生产力透视和语音识别的用户友好界面更加进一步的发展的悟性软件是必要的。
编号：227446 && 大小：283.50KB && 格式：DOC && 上传时间：
关&键&词： 通信 通讯 电子 工程 毕业论文 外文 翻译 中英文 文献
温馨提示：
1: 本站所有资源如无特殊说明，都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。2: 本站的文档不包含任何第三方提供的附件图纸等，如果需要附件，请联系上传者。文件的所有权益归上传用户所有。3.本站RAR压缩包中若带图纸，网页内容里面会有图纸预览，若没有图纸预览就没有图纸。4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。5. 人人文库网仅提供交流平台，并不能对任何下载内容负责。6. 下载文件中如有侵权或不适当内容，请与我们联系，我们立即纠正。7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。
& 人人文库网所有资源均是用户自行上传分享，仅供网友学习交流，未经上传用户书面授权，请勿作他用。
当前资源信息
浏览：1168次
官方联系方式
客服手机：&&& 1:&&&
2:不支持迅雷下载,请使用浏览器下载&&&
3:不支持QQ浏览器下载,请用其他浏览器&&&
4:下载后的文档和图纸-无水印&&&
5:文档经过压缩，下载后原文更清晰&&&
&& && && && && && && && &&
copyright@
人人文库网网站版权所有 苏ICP备号-5}