摘要 |
PURPOSE:To effectively select a phoneme standard pattern and to improve the voice recognition rate by using a phoneme environment table for selecting one which is matched in phoneme environment when selecting the standard pattern of a certain phoneme. CONSTITUTION:The device is equipped with a table 1-7 wherein standard patterns of voice segments and front-rear environment of voice segments that they represent are described and a dictionary 1-6 which has indexes of the voice segments. When a voice is recognized, pattern recognition is performed by using the standard segment of a voice segment selected according to the front-rear environment information of the voice segment described in the table 1-7. Namely, the table 1-7 which contains information showing which position in a word a phoneme is present at and phoneme standard patterns is retrieved by using a phoneme label obtained from the work dictionary 1-6. |