Today Europe is facing larger and more critical language challenges than ever before. The production of multilingual content now far outpaces our ability to translate it by human effort and we must turn to automatic methods to cope. Thus, effective and innovative alternatives must be provided to Europe's citizens and businesses. High performing machine translation technology can be part of the solution.Recent advances in machine translation (MT) technology now show great promise, as systems can be trained automatically from data and achieve respectable performance, even from speech input. However, MT still has very high maintenance costs, and is unsuited to cope with many of today's digital media's relentlessly changing streams of information, across different topics, styles, and genres."Bridges Across the Language Divide" (EU-BRIDGE) proposes to advance speech translation to the point where it can deal with the varying input conditions occurring in digital media, and is able to automatically adapt itself to the changing domains.Training data for modern state-of-the-art statistical speech translation models has become a valuable resource that is difficult and expensive to collect in sufficient quantities for all languages and domains of interest. EU-BRIDGE will therefore seek to reduce the cost of data collection and the dependency on collected data.In its research EU-BRIDGE will address European and non-European languages of high importance to the businesses and citizens of Europe, also dealing with the problem of addressing under-resourced languages and their specific peculiarities.EU-BRIDGE will further be working on European infrastructure and network services in order to provide European businesses with a tangible benefit when operating in a multilingual market, e.g. TV captioning and translation. EU-BRIDGE has identified four use cases which will be serviced by its infrastructure and which will proof the concept of EU-BRIDGE to stimulate use by many more use case developers in the future.