Klaus Ries

Working in the Clarity project on empirical methods for dialogue processing of unrestricted spontaneous speech. The goal of my research is to develop machine learning technology that enables the labeling and segmentation of spontaneous dialogues into meaningful units. We are working on the CallHome Spanish Database.

Previously I have been working in the VERBMOBIL project on statistical language modeling, grammar inference and speech to speech translation, especially Japanese, English, Spanish and German statistical language modeling for speech recognition as well as statistical approaches to dialog modeling. Specialities are adaptive language modeling, word clustering, sequence finding, n-gram and long ranging language models. Additionally I have been working on language modeling for Switchboard and I have been involved in the LVCSR Summer Workshop.

I am writing my thesis on information access to spoken language, proofing that one can use features other than keywords with great effect in this scenario. One can show that dialogue features can be used to differentiate databases, rejoinders in databases and individual segments in a rejoinder. Additional one can show that a rejoinder can be segmented using non-keyword information.

I am working in the Interactive System Labs headed by Alex Waibel. My main affiliations are Universität Karlsruhe (as PhD student) and Carnergie Mellon University, Center for Machine Translation (as visiting researcher and research programmer).

Tee-enabled and Y2K compliant

Use Python today

Klaus Ries