MULTIPLE PRONUNCIATION DICTIONARY STRUCTURING METHOD AND SYSTEM BASED ON THE PSEUDO-MORPHEME FOR SPONTANEOUS SPEECH RECOGNITION AND THE METHOD FOR SPEECH RECOGNITION BY USING THE STRUCTURING SYSTEM

摘要

The performance of the invention by forming the frequently appearing pronunciation variations doctor morphological-based multiple pronunciation dictionary the receiving extends to the representative words in conversational speech and configuring the language model and the vocabulary dictionary using only represent vocabulary, conversational continuous speech recognition multi pronunciation to get improved and standardized output pattern relates to a method and system of pre-built and conversational speech recognition method using the same.; The present invention compares with performing a representative negative text corpus and Allophone extracting the text corpus, respectively, representing sound and Allophone doctor stemming and tagging for each text corpus from the conversational text corpus, tagging results by Eojeol doctor representative of morphological unit negative / extracting Allophone pair and, further, representing sound generating a lexical dictionary from the doctor morphological tagging results of only the representative sound corpus, representing sound vocabulary dictionary and represent negative / Allophone pairs extracted multiple pronunciation through the results a step of generating the dictionary and language models representing negative.