Second European Conference on Speech Communication and Technology

Genova, Italy
September 24-26, 1991


1. 0 TANGORA - A Large Vocabulary Speech Recognition System for Five Languages

Helene Cerf-Danon (1), Steven DeGennaro (2), Marco Ferretti (3), Jorge Gonzalez (4), Eric Keppel (5)

(1) IBM France Paris Scientific Center
(2) IBM USA TJW Research Center
(3) IBM Italy Rome Scientific Center
(4) IBM Spain Natural Language Technology Center
(5) IBM Germany Heidelberg Scientific Center

TANGORA is the prototype of a large vocabulary isolated word recognizer for English developed by IBM Research, USA. IBM research groups in Europe completed experimental Tangora systems for Italian, French, German, and Spanish. Starting from these prototypes, the groups are developing jointly a multi-lingual Tangora integrated in a completely re-designed architecture. It aims at providing a flexible and versatile speech recognizer allowing for running real-life dictation experiments. This paper presents details of the Tangora architecture, problems of the language adaptations and describes the specific issues that have been solved in order to cope with the peculiarities of each language.

Full Paper

Bibliographic reference.  Cerf-Danon, Helene / DeGennaro, Steven / Ferretti, Marco / Gonzalez, Jorge / Keppel, Eric (1991): "1. 0 TANGORA - a large vocabulary speech recognition system for five languages", In EUROSPEECH-1991, 183-192.