Deliverable D 1.1 Deliverable title: Benchmark Project acronym: Translate4Rail Starting date: 01/12/2019 Duration (in months): 24 Call (part) identifier: H2020-S2RJU-OC-2019 Grant agreement no: 881779 Due date of deliverable: M2 Actual submission date: 31-01-2019 Responsible: RNE Dissemination level: PU Status: Final Reviewed: yes GA 881779 P a g e 1 | 53 Document history Revision Date Description 1 31-01-2019 First issue Report contributors Name Beneficiary Short Details of contribution Name RailNetEurope RNE Consolidation of the benchmarking Assessment and analysis Union Internationale Initial benchmark research UIC des Chemins de fer Consolidation of the report GA 881779 P a g e 2 | 53 Table of Contents 1. Executive Summary ....................................................................................................................... 5 2. Abbreviations and acronyms ........................................................................................................ 6 3. Background ................................................................................................................................... 8 4. Objective/Aim ............................................................................................................................. 10 5. TRANSLATION FOR BREAKING LANGUAGE BARRIERS IN THE RAILWAY FIELD ........................... 11 5.1. Translate4Rail project objectives and benefits ................................................................. 11 6. BENCHMARK IN RAILWAYS AND OTHER INDUSTRIES ................................................................ 12 6.1. Railways ............................................................................................................................. 12 6.1.1. Use of speech recognition in simulators ........................................................................... 12 6.1.2. RNE and UIC ongoing projects .......................................................................................... 14 6.2. Air traffic management ..................................................................................................... 15 6.2.1. Activities in the field of speech recognition in air traffic .................................................. 15 6.2.2. MALORCA .......................................................................................................................... 17 6.2.3. AcListant ............................................................................................................................ 19 6.3. Maritime traffic ................................................................................................................. 20 6.3.1. Activities in the field of speech recognition in maritime traffic ....................................... 20 6.3.2. Fraunhofer IDMT: Improved Maritime Communication .................................................. 21 6.4. Military applications.......................................................................................................... 22 6.4.1. Activities in the field of speech recognition in military applications ................................ 22 6.4.2. MFLTS Project ................................................................................................................... 23 6.5. Automotive ....................................................................................................................... 24 6.5.1. Activities in the field of speech recognition in Automotive ............................................. 24 6.6. General Language projects ............................................................................................... 25 6.6.1. EMIME ............................................................................................................................... 25 6.6.2. TC STAR ............................................................................................................................. 26 6.7. The Benchmark of the industries ...................................................................................... 27 7. KEY SOLUTION PROVIDERS ......................................................................................................... 29 7.1. Key aspects for a language solution.................................................................................. 29 7.2. Google Translate ............................................................................................................... 31 7.3. Microsoft ........................................................................................................................... 32 7.4. IBM .................................................................................................................................... 34 7.5. Baidu ................................................................................................................................. 36 GA 881779 P a g e 3 | 53 7.6. Carnegie Mellon University ............................................................................................... 36 7.7. Systran ............................................................................................................................... 37 7.8. DeepL ................................................................................................................................ 38 7.9. iTranslate ........................................................................................................................... 39 7.10. SDL Government ............................................................................................................ 40 7.11. KantanMT ...................................................................................................................... 42 7.12. List of speech to speech applications ............................................................................ 43 7.13. List of translation devices .............................................................................................. 46 7.14. The Benchmark of the Translation tool and providers ................................................. 48 8. CONCLUSIONS ............................................................................................................................. 51 9. References .................................................................................................................................. 52 GA 881779 P a g e 4 | 53 1. Executive Summary Benchmark – Speech to Speech: State of the Art (Internet-based search) The concept of the Translate4Rail project is to offer drivers a fully comprehensive set of predefined standardised messages which encompass all they have to exchange with an infrastructure manager traffic controller in normal or exceptional operational situations in a country where they do not understand nor speak the local language. An IT tool will then be implemented to enable the driver and the traffic controller to understand each other even though each of them speaks in his/her native language. This will capitalise on the works already committed on this matter between Infrastructure Managers (IMs) and Railway Undertakings (RUs) at RNE and UIC level. These works have dealt with the analysis of the various types of operational situations needing exchanges between RUs drivers and IMs signallers. The project will enable to test these works and to further harmonise and standardise pre-defined messages in the light of the tests carried out. Such messages will be uttered by the driver or the traffic controller. They will then be identified, translated and uttered in the language of the other party by the given tool. The project will define the functional characteristics of the tool which will create a frame for the exchanges between drivers and traffic controllers. This tool will use voice recognition and translation applications. The tool will be tested on pilot trains running on cross border sections of rail freight corridors where drivers have to use different languages. The project intends to at least maintain the level of safety, increase the traffic fluidity at borders and to increase the competitiveness of the rail sector. Years ago, research has been started regarding the use of computers to recognize and render sign language. Currently, there are numbers of research projects focused also for text and speech automatic translation. Technology is rapidly changing and improving the way we know the world operates today and could in the future. A typical speech-to-speech translation system consists of three components: speech recognition, machine translation and speech synthesis. Many techniques have been proposed for the integration of speech recognition and machine translation. However, corresponding techniques have not yet been considered for speech synthesis. [1] Now, we are dealing with the question: What are the areas in which you see artificial intelligence playing a role? And there is a simple answer: Whether we admit it or not, new technologies and artificial intelligence in the future, perhaps soon, will greatly affect our work and our way of life. It is up to us how we can take advantage of the opportunities that could positively influence our daily lives. For man, language is the most natural thing, and so it is the greatest challenge that machines face, just like humans. Depends on the topic, this benchmark report is partly a compilation of internet-based search. GA 881779 P a g e 5 | 53 2. Abbreviations and acronyms Abbreviation / Acronyms Description