Insights into the Dialogue Processing of VERBMOBIL

TL;DR

This paper describes the dialogue processing component of VERBMOBIL, a speech-to-speech translation system, emphasizing a modular approach that combines multiple robust components to improve translation accuracy.

Contribution

It introduces a multi-component dialogue processing method for speech translation systems, demonstrating its effectiveness on real data and in improving German to English translation.

Findings

Effective dialogue processing on real data

Contributes to correct German-English translation

Uses a combination of simple, efficient components

Abstract

We present the dialogue module of the speech-to-speech translation system VERBMOBIL. We follow the approach that the solution to dialogue processing in a mediating scenario can not depend on a single constrained processing tool, but on a combination of several simple, efficient, and robust components. We show how our solution to dialogue processing works when applied to real data, and give some examples where our module contributes to the correct translation from German to English.