| 摘要 |
PROBLEM TO BE SOLVED: To provide a statistic language model generating device for generating a statistic language model which improves recognizing performance in voice recognition and increases the appearance probability of a word which is easily included in speech contents and also to provide a statistic language model generating program. SOLUTION: The statistic language model generating device is provided with a voice recognizing means 7 for recognizing nearest text data and a large amount of past text data as voice, a storage means 9 for storing nearest text data (integrating means 5), the large amount of past text data (integrating means 3) and a recognition result recognized by the voice recognizing means 7, a probability weight calculating means for calculating the first probability weight of n-gram based on the large amount of past text data, calculating the voice probability weight of n-gram based on nearest text data and calculating the third probability weight of n-gram based on the recognition result and a language model generating means 11 for generating the statistic language model based on the first, second and third probability weights. |
