摘要 |
PURPOSE:To prevent a voice recognizing performance from deteriorating even when an honorific title is added by storing a standard voice pattern to the voice for the honorific title only into a standard voice pattern memory, deleting the part to the honorific title and comparing an input voice pattern and a standard voice pattern. CONSTITUTION:When a voice '-san(mr.-)' is produced, an input voice pattern to the voice '-san' is prepared in a pattern preparing part 4 and inputted to an address from deleting part 6. In the honorific title deleting part 6, the coincident part of the part to the 'san(mr.)' of the input voice pattern and the standard address from voice pattern to the 'san' in a standard address from voice pattern memory 5 is deleted and stored into a standard voice pattern memory 7. When the voice '-san(mr.)' is produced, the honorific title deleting voice pattern is prepared in which the part to the honorific title 'san(mr.)' is deleted in the address from deleting part 6, inputted to an identifying part 8 and compared with the standard voice pattern in the standard voice pattern memory 7 in the identifying part 8.
|