Method for voice searching in text of an expression element of a request by segmenting voice request is in second continuation of discreet phonetic units, for searching occurrences of expression object of request in text

摘要

A text is transcribed in one first continuous phonetic unit. The voice request is segmented in second continuation of discreet phonetic units, for searching occurrences of an expression object of the request in the text by aligning of the first and second continuous phonetic units. The unit of acoustic-phonetic decoding uses an algorithm of hidden Markov model.