PURPOSE: To improve the speech recognition rate of a speech recognizing device by using a speech recognizing dictionary to a person who always inputs voices at the time of speech recognition.
CONSTITUTION: Plural speech recognizing dictionaries prepared in accordance with classified voice features are previously stored in a dictionary storing means 5. At the time of speech recognition, for example, an operator code or operator card is inputted or an voice analyzer 3 analyzes and recognizes the feature of the voice inputted by an operator, and then, one speech recognizing dictionary suitable to the operator who inputs the voice is selected from the dictionaries stored in the storing means 5 by means of a selecting means 8. Then matching is performed between the selected speech recognizing dictionary and the voice signal obtained from a voice converting means 2 and a recognized result is outputted. Therefore, the optimum speech recognizing dictionary is used for an operator even if his voice quality changes, resulting in improvement in speech recognition rate.
JPS62134697 | JAPANESE LANGUAGE PROCESSOR |
WO/2023/091158 | EPHEMERAL LEARNING OF MACHINE LEARNING MODEL(S) |
JPS60212799 | VOICE RECOGNITION EQUIPMENT |
IZUMI SANAE
Next Patent: AUTOMATIC OUTPUT SWITCHING DEVICE FOR VOICE SYNTHESIS