"Muu" projekt AR13060
AR13060 "Kõnetuvastussüsteem meditsiinirakendustele (1.02.2013−31.08.2015)", Ivo Fridolin, Tallinna Tehnikaülikool, TTÜ Tehnomeedikum.
3.2.1201.13-0010
AR13060
Kõnetuvastussüsteem meditsiinirakendustele
Estonian Speech Recognition System for Medical Applications
1.02.2013
31.08.2015
Teadus- ja arendusprojekt
Muu
ETIS klassifikaatorAlamvaldkondCERCS klassifikaatorFrascati Manual’i klassifikaatorProtsent
4. Loodusteadused ja tehnika4.7. Info- ja kommunikatsioonitehnoloogiaT121 Signaalitöötlus 2.2. Elektroenergeetika, elektroonika (elektroenergeetika, elektroonika, sidetehnika, arvutitehnika ja teised seotud teadused)100,0
AsutusRiikTüüp
SA Archimedes
PerioodSumma
01.02.2013−31.08.201537 774,69 EUR
37 774,69 EUR
Riiklik programm Info- ja kommunikatsioonitehnoloogia alase teadus- ja arendustegevuse toetamine

The goal of the project is to develop a novel prototype of an Estonian speech recognition system for user interfaces of medical applications. The system exploits language technology and methods for adapting speech recognition to different users and medical domains. Automatic speech recognition (ASR) technology has been available for “big” languages (like English, French, German, etc.) already for a couple of decades. The ASR-applications include dictation and editing of electronic documents by voice (e.g. Dragon Naturally Speaking), automatic call centers, voice controlled interfaces, assistive technology for handicaps, medical dictation/transcription systems (e.g. Dragon Medical, MediSpeech). Radiology is one of the medical domains where ASR has been successfully implemented replacing the traditional keyboard based text entry. Comparing with the widely used approach where reports dictated by radiologists are first recorded and then converted to text by human transcribers, the use of ASR technology results in up to 50% faster and less expensive report delivery. In the current project we will:  exploit the recent achievements in Estonian ASR and adapt the prototype of Estonian large vocabulary continuous speech recognition for the medical domain;  study and implement the methods for user and domain specific adaptation;  as a final result, provide a prototype of Estonian speech recognition system that can be integrated into user interfaces of different applications in healthcare domains (radiology, pathology) in all hospitals in Estonia. Radiology Centre at North Estonia Medical Centre (NEMC), being technologically the most advanced radiology unit in Estonia, has been chosen as a test environment for piloting the project applicability. The project is in good compliance with the sub-measure goals since language technology as a part of ICT, is one of the key components in building new innovative user interfaces and solutions providing better access to different e-services. In Estonia, language technology has a strategic role in enabling the use of Estonian language in the information society and in longer term, it supports the preservation of Estonian language in the future world. The Development Plan of Information Society 20131 states the importance of sustainability of Estonian language and culture in information society; this can be achieved only by stronger support for research and development of Estonian language technology and by promoting the use of language technology tools in different ICT-systems. Estonian language technology has achieved a strong position in Estonian ICT Survey EST_IT@20182 carried out in 2008, and due to specific funding scheme – the National Program for Estonian Language Technology (NPELT) 2006-2010 and 2011-2017 – it has made a further progress in recent years. The progress has been especially remarkable in the field of automatic speech recognition resulting in several demo-applications, e.g. Transcribed Speech Archive Browser (http://bark.phon.ioc.ee/tsab/p/index) and apps for Android platform. With our current project application we make the first attempt to implement automatic speech recognition in
The goal of the project is to develop a novel prototype of an Estonian speech recognition system for user interfaces of medical applications. The system exploits language technology and methods for adapting speech recognition to different users and medical domains. Automatic speech recognition (ASR) technology has been available for “big” languages (like English, French, German, etc.) already for a couple of decades. The ASR-applications include dictation and editing of electronic documents by voice (e.g. Dragon Naturally Speaking), automatic call centers, voice controlled interfaces, assistive technology for handicaps, medical dictation/transcription systems (e.g. Dragon Medical, MediSpeech). Radiology is one of the medical domains where ASR has been successfully implemented replacing the traditional keyboard based text entry. Comparing with the widely used approach where reports dictated by radiologists are first recorded and then converted to text by human transcribers, the use of ASR technology results in up to 50% faster and less expensive report delivery. In the current project we will:  exploit the recent achievements in Estonian ASR and adapt the prototype of Estonian large vocabulary continuous speech recognition for the medical domain;  study and implement the methods for user and domain specific adaptation;  as a final result, provide a prototype of Estonian speech recognition system that can be integrated into user interfaces of different applications in healthcare domains (radiology, pathology) in all hospitals in Estonia. Radiology Centre at North Estonia Medical Centre (NEMC), being technologically the most advanced radiology unit in Estonia, has been chosen as a test environment for piloting the project applicability. The project is in good compliance with the sub-measure goals since language technology as a part of ICT, is one of the key components in building new innovative user interfaces and solutions providing better access to different e-services. In Estonia, language technology has a strategic role in enabling the use of Estonian language in the information society and in longer term, it supports the preservation of Estonian language in the future world. The Development Plan of Information Society 20131 states the importance of sustainability of Estonian language and culture in information society; this can be achieved only by stronger support for research and development of Estonian language technology and by promoting the use of language technology tools in different ICT-systems. Estonian language technology has achieved a strong position in Estonian ICT Survey EST_IT@20182 carried out in 2008, and due to specific funding scheme – the National Program for Estonian Language Technology (NPELT) 2006-2010 and 2011-2017 – it has made a further progress in recent years. The progress has been especially remarkable in the field of automatic speech recognition resulting in several demo-applications, e.g. Transcribed Speech Archive Browser (http://bark.phon.ioc.ee/tsab/p/index) and apps for Android platform. With our current project application we make the first attempt to implement automatic speech recognition in
KirjeldusProtsent
Rakendusuuring45,0
Alusuuring20,0
Katse- ja arendustöö35,0