Title: Language Models beyond Word Strings

Abstract: In this paper we want to show how n-gram language models can be used to provide additional information in automatic speech understanding systems beyond the pure word chain. This becomes important, when the spoken utterance can only be interpreted in the context of the dialogue history. We show how n-grams can (1) help to classify prosodic events like boundaries, accents, and sentence mood, (2) be extended to directly provide boundary information in the speech recognition phase, (3) help to process speech repairs, and (4) detect and semantically classify out-of-vocabulary words. The approaches can work on the best word chain or a word hypotheses graph. Examples and experimental results are provided from our own research within the EVAR information retrieval and the VERBMOBIL speech-to-speech translation system.

Curriculum: Elmar Nöth, obtained his `Diplom' in Computer Science and his doctoral degree at the University of Erlangen-Nürnberg in 1985 and 1990, respectively. From 1985 to 1990 he was a member of the research staff of the Institute for Pattern Recognition (Lehrstuhl für Informatik 5), working on the use of prosodic information in automatic speech understanding. Since 1990 he is an assistant professor and the head of the speech group at the same institute. From October 1992 until March 1993 he was a Substitute Professor for Phonetics and Phonology at the University of Stuttgart, Germany. His current research activities concern prosody, the linguistic processing and knowledge representation in automatic speech understanding systems. During 1979-1980 he spent his `junior year abroad' at the Massachusetts Institute of Technology doing research in computer vision. In September and October 1993 he was a visiting scientist at the `Centre de Recherche Informatique de Montreal' (CRIM) in Canada with Prof. de Mori. He is the author or coauthor of one book and about 150 technical articles. He is a member of GI and ISCA.

Noeth.ps Noeth.pdf