JPS59116799A - Dictionary checking system for voice recognition equipment - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

Detailed Description of the Invention (1) Object of the Invention The present invention relates to a method for checking a speech standard pattern file (dictionary) in a speech recognition device. was not checked. Therefore, if someone mistakenly uses another person's dictionary, not only will it not be possible to recognize it properly, but there is a risk that the other person's standard speech pattern stored in the dictionary may be destroyed by re-registering the dictionary. (3) Implementation of the invention EXAMPLE An example of the present invention will be explained based on the drawings. I will explain what happens when creating a shortcut dictionary. A speaker who wishes to create a dictionary inputs his or her name into the voice input section 1 shown in the figure. The input voice signal is sent to the voice recognition section 2 shown in the figure, converted into a standard voice pattern, and then sent to the control section 3, where it is registered in a predetermined area 41 of its internal memory 4. Next, by inputting a single syllable, word, etc. to be registered into the voice input section and inputting the characters (code) of the syllable or word from a keyboard or the like, the name is registered in the control section. After registration is complete, save the registered pattern to external memory 1. A dictionary is created by storing in -. The voice standard pattern of the first registered name becomes the iD of the dictionary.

Next, the operation during recognition will be explained. First, a dictionary floppy disk is set in the 70 floppy disk device 5. The speech standard pattern of the name, which is an ID, from the set dictionary floppy is sent to the control section and further sent to the speech recognition section. When the speaker reversely inputs his or her name into the voice input section, the speech pattern to be recognized is sent to the speech recognition section, where it is matched with the standard speech pattern of the name sent from the control section, and the speech pattern is sent to the dictionary. will be checked.

If a match is found, the dictionary becomes the dictionary of that speaker, and the contents of the floppy disk can be transferred to the internal memory 4 and normal recognition of monosyllables, singletons, etc. can be started.

However, if a match cannot be made, the answer l is not the speaker's dictionary, and the speaker is notified of this by displaying this on the display 6 or by outputting audio from the audio output unit 7. However, it does not move to recognition operation.

In normal recognition, the speech pattern to be recognized is compared with all patterns in the dictionary and the word with the highest degree of matching is selected as the recognition result, but in the above iD recognition, only the pattern of the name is compared, It is assumed that there is a match only when the degree of matching takes a value equal to or higher than a predetermined value.

(4) Effects of the invention As mentioned above, in order to check whether the dictionary is the speaker's dictionary, another person's dictionary is used to destroy the other person's standard speech patterns stored in the dictionary. There is no.

It is also possible to use a number as an ID to match the dictionary and the speaker, but in that case, the ID may be forgotten and the dictionary cannot be used, or the wrong number is memorized and someone else's dictionary is used. However, since the iD is used as the name, it is thought that there will be no forgetting or misremembering of i1).

[Brief explanation of the drawing]

The figure is a block diagram of one embodiment of the present invention, in which 1 is a voice input section, 2 is a voice recognition section, 3 is a control section, 4 is an internal memory, and 4 is a block diagram of an embodiment of the present invention.
1 is an iD area, 5 is a floppy disk device, 6 is a display, and 7 is an audio output section.

Claims (1)

[Claims] In a specific speaker speech recognition device that registers a speech standard pattern in advance and performs recognition by comparing the speech pattern to be recognized with the speech standard pattern, the name of the creator of the speech standard pattern at the time of registering the speech standard pattern. The voice pattern of the name is also registered at the same time, and the voice pattern of the name registered as the ID of the voice standard pattern file created by storing the registered pattern in an external storage medium is used. A dictionary check method that checks whether the dictionary being used is the dictionary of the speaker by matching the pattern with the voice pattern of the name used as the ID of the dictionary.