170 likes | 281 Views
A magyar beszédtechnológia helyzete és távlatai (Status Report of Hungarian Speech Technology). Németh Géza BME Távközlési és Médiainfor matikai Tanszék Beszédtechnológiai Laboratórium Budapest University of Technology & Economics Department of Telecommunications & Media Informatics
E N D
A magyar beszédtechnológia helyzete és távlatai(Status Report of Hungarian Speech Technology) Németh Géza BME Távközlési és Médiainformatikai Tanszék Beszédtechnológiai Laboratórium Budapest University of Technology & Economics Department of Telecommunications & Media Informatics Speech Technology Laboratory nemeth@tmit.bme.hu
Overview • What is it? • Why is it important in general? • Why is it important in Hungary? • History • Recent results • Available resources • Research challenges • Application challenges
What is it? Artificial replacement of any element of the human speech chain Relyon … mathematics, information technology, physics, neurology, linguistics, psychology and electrical engineering [http://www.hlt-platform.hu/en/the_definition_of_speech_and_language_technology.html}
Why is it important in general? • Language <> text • Speech is the main modality of the expression of language • It is the most efficient • Disadvantage of loss of speech vs. loss of sight • In some contexts (in-car, manufaturing, …) preferred communication channel • Big data source (natural, real, …)
Why is it important in general? Related to speech technology [Gartner hype-cycle on Emerging technologies July 2012]
Why is it important in Hungary? • Wehave a uniquelanguage (agglitunative, free wordorder) • Extra effort - Middle-sized market (73rd intheworld[Ethnologue]) • Multinationalsgettinginterested (Google, Nuance, …) but • Tailor-made, highqualitysolutionscost toomuch <> justsufficienteffort • Prominens résztvevők • Maróth Miklós (alelnök, MTA, nyelvész); • Gróh Gáspár (Áder János köztársasági elnök megbízásából, közíró); • Kelemen Csaba (fővh, ICT fejlesztés, Németh Lászlóné miniszter köszöntője, NFM); • Csizmadia Norbert (tervezéskoordinációért felelős államtitkár, NGM); • L. Simon László (kultúráért felelős államtitkár, EMMI); • Hoffmann Rózsa (oktatásért felelős államtitkár, EMMI) írásos köszöntője; • Bába Iván (közigazgatási ügyekért felelős államtitkár, KülügyM); • Korányi László (kül- és belkapcsolati elnökhelyettes, villamosmérnök, NIH)
History of vehicle and speech technology • 1791 • 2012
Recent real-life results of of Hungarian speech technology MailMondó Westel BME TMIT 1999 T-Mobile Freedom BME TMIT 2002 Scientific Informatika a Látássérültekért Westel BME TMIT 2003 T-Mobile MIT Systems Digital Natives BME TMIT 2008 AITIA MonSpeech Vodafone Montana, AITIA, 2012 BME TMIT, MTA Nytud
Available resources • World-class language and speech technology co-operative R&D know-how • www.hlt-platform.hu • SMEs (AITIA, Morphologic, Nextent, … ) • International networks • Lack of large industrial R&D centers • Lack of focused attention, quality requirements META-NET
Research challenges 1 • Accurate reference speech processing infrastructure • Processing of spontaneous interactions • Collecting and labelling enough (?) data • Unfunded international efforts (e.g. U-STAR) • Rule-data driven combination • Cognitive Infocommunications • Cognitive Robotics • Eto – communications • Just ripe applications
Research challenges 2 How to avoid the „uncanny valley”
Application challenges 1 • 62% of 15-69 yearHungarianpopulation is internet user • Whataboutthe rest (38%)? • Equalaccesstoinformation??? • Speechtechnologymayhelp (magyarorszag.hu, 112, MÁV, BKV, Volán) • Example: www.gyogyszervonal.hu, www.metnet.hu • Disabilityapplications • Screenreadersforthevisuallyimpaired • Electronicacesstoteaching and otherwrittenmaterial • Example: www.robobraille.org, VoxAid
Application challenges 2 • Speech technology in education • Games for kindergarten and schoolchildren • Example: GOH hearing screeing at 3 years • Interactive multimodal teaching material • Motivation of Hungarian kids in minority situation • Rehabilitation of aphasia, autism, problems…
Application challenges 3 • Speechtechnologyinthehealthindustry • Automation of operations (instructions, notetaking) • Automation of findingsdictation • Earlydiagnosis and rehabilitation of larynxproblems, depression, etc. byvoice • Remotehealthapplications (e.g. warningaboutmedication, windowclosure, etc.) • Supervision of dementia, Alzheimer, …
Application challenges 4 • Speech technology in the content industry • Interdisciplinary integration • Speech technology – medical education – social workers (IBM – Hungarian government?) • Digital public education and intelligent home program (Microsoft – Hungarian government?) • Multi-model content analytics (polls??) • Banks, retail industry information services • Car infotainment (Audi, Daimler – Hungarian gov?) • Speech controlled home • Smartphone, smartTV • Smart washing machine, ……
Application challenges 5 • Speechtechnologyinmanufacturing • Warehouseautomation • Productionwarning • Speechinstructions • Talkinguser manuals • 3DICC 3D Internet Based Control and Communication
Mélyebb érdeklődőknek: http://speechlab.tmit.bme.hu/ http://magyarbeszed.tmit.bme.hu/ Köszönjük az támogatását. (Teleauto, BelAmi, EtoCom -TÁMOP-4.2.2-08/1/KMR-2008-0007- , BME Kutatóegyetemi -TÁMOP-4.2.1/B-09/1/KMR-2010-0002- , CIP CESAR, AAL PAELIFE projektek) Hozzászólások (Comments, questions)