THE BBN BYBLOS 1997 LARGE VOCABULARY CONVERSATIONAL SPEECH RECOGNITION SYSTEM

This paper presents the 1997 BBN Byblos Large Vo-cabulary Speech Recognition (LVCSR) system. We give an outline of the algorithms and procedures used to train the system, describe the recognizer configuration and present the major technological innovations that lead to performance improvements. The major testbed we present our results for is the Switchboard Corpus, where current word error rates vary from 27 % to 34 % depending on the test set. In addi-tion, we present results on the CallHome Spanish and Ara-bic tests, where we demonstrate that technology developed on English Corpora is very much portable to other problems and languages