DFKI-LT - Verbmobil
Translation of Spontaneous Speech
Verbmobil is a long-term interdisciplinary Language Technology project. The Verbmobil System recognizes spoken input, analyzes and translates it, and utters the translation.This speaker-independent system processes spontaneous speech. Verbmobil offers assistance in multilingual dialog situations in certain domains (e.g. scheduling appointments, travel planning and making hotel reservations). The project is a joint initiative involving information-technology companies, universities, and research centers. It has been divided into two phases, which are both funded by the German Federal Ministry for Education, Science, Research and Technology (BMBF) and additionally financed by the industrial partners.
The first fully integrated system, the Verbmobil Demonstrator, was unveiled to the public during the CeBit in 1995 by Dr. Jürgen Rüttgers, the German Federal Minister for Research (BMBF). The Verbmobil Demonstrator recognizes spoken German utterances within the context of appointment negotiation (vocabulary: 1292 words), analyzes them, and translates them into spoken English. The Verbmobil Research Prototype (vocabulary: 2500 words) was presented at CeBIT 1997. It additionally recognizes Japanese utterances (vocabulary: 400 words) and translates them into English.
In phase 2, the Verbmobil System will translate spontaneously-spoken dialogues robustly and bidirectionally for German/English and German/Japanese. To achieve a relatively broad coverage of cooperative colloquial speech and various accents, Verbmobil will be speaker-independent with the ability to process larger vocabularies (about 10,000 words) for German and English. At the end of the second phase, a system will be available on a central server that is no longer dependent on a close-speaking microphone. The mobile accesss of the Verbmobil speech server is made possible by using digital cellular phones.
- Speeker independent recognition
- Knowledge based language processing
- Semantic and statistic translation
- Context sensitive dialogue processing
- Multilingual generation and speech synthesis
|Funded by:||Federal Ministry of Education and Research|
|Project Manager:||Wolfgang Wahlster (Wolfgang.Wahlster@dfki.de)|
|Contact:||Reinhard Karger (Reinhard.Karger@dfki.de)|
|Duration:||01.01.1993 - 31.12.2000|