Speech is the most important natural human communication form. Thus, speech recognition is the most natural interface for applications and allows development of applications that enable human communication across language and culture barriers.
In our research, we are concentrating on techniques and algorithms for improving speech and text applications that are robust in multilingual environments. This includes the rapid development of language recognizers for new tasks and languages – here, it is necessary to significantly reduce the time and costs required.
We believe that this is an essential prerequisite for making voice-driven applications attractive to the user and for a further spread of these applications in speech areas for which there are only few or no resources at present.