Hello my name is Alex Waibel. I'm a professor at Carnegie Mellon University and at Karlsruhe Institute of Technology, the area of research that I've been engaged in for many, many years is language processing in the widest sense, in that we are interested in building systems that allow people to communicate with each other, regardless of culture and language barriers. One of the core problems of this vision is to build machine translation systems, systems that can translate from one language to another. The goal of the smoke is therefore to introduce you to the techniques and the technologies of automatic translation by machine. All of this leads us to a vision that really make translation so ubiquitous that language barriers go away all together, language barriers appear in various forms and various modalities, and we will address those because they complicate the translation problem. Let us consider this vision of the future in this picture, you see various ways in which we can deploy machine translation and in different situations in which people require such communication with each other. The upper picture on the right, you see a situation where people want to talk with each other in a dialogue. In this situation very often. You would like to have consecutive translation, which means that one person speaks something in one language, it gets translated, and the other person hears it in their language, and then after that, they can react in their language and translate something back to the first person. This consecutive translation, of course, takes more time, because you need the translation to complete before the other person can react. Another way of doing this is so - called simultaneous translation, which you see in the lower right picture here. What we have is a situation where speaker is producing a monologue for a continuous period. In this case, we would want to have translation simultaneously, while the person is speaking. In this case, of course, we can't wait for the translation to complete before the person says the next sentence here. What happens is translation is done simultaneously, while the speaker is still speaking and producing more sentences as input. Now, if we put this all together, we might actually eventually have a vision where we have a situation where multiple people can speak in multiple languages and Indians or interact with each other, regardless of their language and their culture. If you consider the picture on the lower left. Imagine a meeting with several people speaking in their respective languages and then interacting freely back and forth. Here we may need consecutive translations, simultaneous translation. We may need different forms of deployment. F- different forms of input and output. It will look and consider these various possibilities as well