Speech recognition device implementing a syntactic permutation rule using an audio processor for acquisition of an audio signal and a linguistic device for decoding

摘要

A processor (2) digitizes an audio signal from a microphone (3) supplied by a signal acquisition circuit (4) and translates digital samples into acoustic symbols chosen from a predetermined alphabet, while a linguistic decoder (6) uses an acoustic model (7) and a language model (8) implemented by a hypothesis-based search algorithm (9). The language model is used to submit hypotheses to the search algorithm, determining the most probable sequence of words and then the most probable solution at the end of a sentence.