Language and Pronunciation Modeling in the CMU 1996 Hub 4 Evaluation
Kristie Seymore, Stanley F. Chen, Maxine Eskénazi · 1999
We describe several language and pronunciation modeling techniques that were applied to the 1996 Hub 4 Broadcast News transcription task. These include topic adaptation, the use of remote corpora, vocabulary size optimization, n-gram cutoff optimization, modeling of spontaneous speech, handling of unknown linguistic boundaries, higher order n-grams, weight optimization in rescoring, and lexical modeling of phrases and acronyms. 1. INTRODUCTION The language modeling component of the CMU 1996 Hub 4 system was developedthrougha series of experiments in topic adaptation,the use of remote corpora, vocabulary size optimization, n-gram cutoff optimization, modeling of spontaneousspeech, handling of unknown linguistic boundaries, higher order n-grams, weight optimization in rescoring, and lexical modeling of phrases and acronyms. These experiments were carried out using the Sphinx-III speechrecognition system: one language model was used with the Sphinx-III decoder to generate N-best lists f...