SPEECH RECOGNITION DEVICE

摘要

A speech recognition device wherein while adequately limiting the vocabulary to be recognized and collected from outside the speech recognition device, the uncomfortable feeling of the user due to the limitation can be eliminated. The speech recognition device comprises a language model storage section (104) storing a language model representing the occurrence probability of each word or word string, an acoustic feature value extracting section (101) and a collating section (102) both for extracting the feature value of an inputted speech signal, collating the extracted feature value with the language model stored in the language model storage section (104), recognizing the word or word string corresponding to the speech signal, and outputting the recognized word or word string as a speech recognition result, an acquired word signal receiving/analyzing section (105) for acquiring and analyzing the word, and a language model adjusting section (110) for determining the occurrence probability of the word according the passage time from the acquisition of the word by the acquired word signal receiving/analyzing section (105), reflecting the determined occurrence probability on the language model stored in the language model storage section(104), and adjusting the language model.