JPS58155451A - Voice inputting system - Google Patents

Abstract

PURPOSE:To execute user's confirmation in advance, and to obstruct utilization by other person, by deciding and registering a secret message which is different in each individual of plural users, and collating it with a voice standard pattern to its message when said message is used. CONSTITUTION:A speaker A inputs a secret message (password) inputted through a telephone set 6, a circuit interface 8, a switch 11, etc., to a secret voice recognizing circuit 12, and the speaker A is recognized by matching of a secret message of the speaker A, and a voice standard pattern of a secret message dictionary 9 in which secret ward of plural speakers are registered. By a discriminating signal of the circuit 12 by this vertification, a dictionary selecting circuit 13 is controlled, command/data of an each registered person command/ data dictionary 15 are inputted to a command/data voice regonizing circuit 14, and the switch 11 is switched. Subsequently, a command inputted from the telephone set 6 by the speaker A is compared with command/data of an individual from the dictionary 15 in order by the circuit 14, its response is made to the speaker A through a selecting and editing circuit 17, etc., and utilization by other person is obstructed.

Description

DETAILED DESCRIPTION OF THE INVENTION (a) Technical Field of the Invention The present invention relates to a voice input method comprising a voice input means and a confirmation means by voice synthesis output.

(b) Prior art and differences At present, input means based on voice recognition are being put into practical use, but most of them rely on voice input methods by specific speakers.

Then, the #It is selected from among the #Its that have been registered in advance.

FIG. 1 shows a conventional example of a voice input method. In the figure, when a specific speaker A inputs a voice input command, for example, a command #'i'"start operation of unit 1", to the voice pattern recognition path 1, the command of the speaker A is recognized in advance.

The self-answers are checked one by one from the Dictionary 2 in which the data-choking quasi-patterns are registered, and the one with the highest degree of matching is selected, and the voice pattern recognition circuit 1 controls the command corresponding to the command. A signal is output, and the control signal instructs the operation of the command target device 3 (namely, the first device).

This command is displayed on display 4 for confirmation. Person A is notified.

As mentioned above, even in the voice input of a specific speaker A,
Since the recognition accuracy of the voice is not 100-, confirmation of the command is required.

When considering the wide use of voice input methods, it is desired that users in a wide area can access them remotely using communication lines. In this case, the following diluted sweat is required.

1) It is possible to expand the number of users from a designated individual to multiple users, reject M3 users, and prevent malicious access.

2) There are no restrictions on the number of terminals that can be used for input.

3) Users can check whether voice input (commands and data) has been correctly standardized.

(C) Purpose of the Invention In order to solve the above-mentioned problems, the present invention provides a new voice input method that verifies the speaker, processes the speaker's input information, and returns the result to the speaker. Providing (d) Structure of @Ming and the purpose of this according to the present invention is as follows: 1) Decide a different password message (password) for each individual for all multiple users, register this, and when using it, The person in question is recognized by comparing the message with a standard voice pattern.

2) A standard voice pattern dictionary for recognizing voice input for commands and data is prepared for each registered user, and when a user is identified in 1), a dictionary corresponding to that user is prepared. Use.

3) - The woven result is sent back to the user 4 in the form of a replay through voice synthesis, and the user 4 is asked to check whether the voice input has been woven correctly.

4) Commands and data are first sent to the command target device from the user's MiK.

5) The operating status of the command target device is achieved by the above-mentioned means, such as notifying the user by voice synthesis.

(・) Examples of the invention The present invention will be described below with reference to the embodiment shown in FIG.

FIG. 2 is a system diagram of voice input. In the same figure,
When the person speaking inputs a secret message (password) into the transmitter 6-1 of the telephone 6, the message is input through the telephone line M7, the line interface 8, and the message circuit 10, where it reaches a predetermined level. The set password is entered through the switch 11 (G) 11-It in the password voice 1 line circuit 12, and is input into the password message dictionary 9 in which the password of Ministry A and the password words of multiple speakers are registered. It is recognized by matching with the standard voice pattern, and when registrant A is confirmed, the password voice (
Password) The recognition circuit 12 outputs a p identification key, and the output opens the gate of the selection circuit 13, leading the command and data dictionary for registrant A to the voice receiving circuit 14, and the switch IIf'i gate. It is switched to 11-2@.

Next, a predetermined command is inputted into the transmitter 6-1 by the speaker, and the command is sent to the telephone 6-' 1111I7-line interface 8-voice receiving circuit 10-switch 11
11 and 2, and is input to the speech recognition circuit 14. As a result, the command and data dictionary 1 for the Ayu Ministry A
5) Sequentially, the content and the voice pattern of the command of the previous message 1iA transmitted are compared, and the one with the highest matching is selected, and the connection of the command is selected from the cough circuit 14 as the input buffer memory 16. It is input to the editing circuit 17.

As a result, the rear f that corresponds to the Mukuro command recognition signal
The phrase is received by the output fWti memory 18, the repetition syllable pattern of the command is performed in the selection editing circuit 17, and the mosquito editing pattern is synthesized with the voice of the command in the voice reproduction circuit 19, and the synthesized voice is bal-interface 8-
The message is received by the telephone line @7-6-2 of the telephone kj6 and confirmed by the speaker A.

If the confirmation of the command is OK, input the next new command or data using the transmitter 6-1.

The input pad 2A1 is accessed through the same route as above for the cultivation command company.
The old commands accumulated in 6 and existing up to that point are sent to the command target device 20.

On the other hand, in the input of the command Zengsei Shaoxian times w & 140,
If Makoto is unable to do so (for example, when an unregistered erroneous command is input), a retransmission request signal is sent to the selection editing circuit 17, the output syllable is edited here, and the audio reproduction circuit 18 requests a retransmission. A voice signal, for example, "Send to -" is sent to the station A to prompt input.

Next, if the songwriter A listens to the replay of Sedo' for confirmation and determines that it is an error, he sends a cancel command, and this is recorded audibly and transferred to the input buffer memory 1.
Delete the contents of 6 and prevent output to the controlled device.

After canceling, the story 4A re-enters the correct command/data.

Next, if the composer A fails to hear the chanting prayer for confirmation, by sending the [1g command again, this will be acknowledged by Zeng P, and the content of the human buffer will be selected and edited. circuit 17
B repeats the previous command/data by sending it to B, the speaker A listens to this, and the rest is repeated in each case (normal ring -/if the command/data is other than cancel cancel and reconfirm introduction). Assuming that the previous command/data has been confirmed as correct by the controller A, the contents of the input pad 7a 16 are output to the controlled device 20.

On the other hand, the operation status signal (including the operation status and data) of the cough device 20 is outputted from the output fTh'l memory by the selection editing circuit 17, and the operation status M signal is edited into the output section '#'. , is synthesized into the operating state fp by the audio reproduction circuit 14, and is output to the receiver 6-2 via the same path as described above.

Effects of the Invention According to the present invention, when inputting voice, the user's 1If
IiIl: Since the 4M- is shown to be the right issue, it is possible to prevent others from using it as an original, and since the command and data input Fi voice is pre-confirmed, it is possible to prevent the control equipment from colliding with each other. Turtle.

Furthermore, these devices have the advantage of being able to partially share the functions of the FI stick for monitoring and measuring information through voice synthesis.

[Brief explanation of the drawing]

FIG. 1 shows a conventional 1i' input system, and FIG. 2 shows an embodiment of the present invention. In the figure, 1 is the audio signal (turn recognition W&circuit), 2 is the back door standard pattern dictionary, 3 is the finger switch, 4 is the display, 5 is the microphone, 6 is the electric wire, and 6-1 is the transmitter. vessel, 6-2 one, 12
1 is a password recognition circuit, 13 is a silent selection circuit, 14 is a command/data sound Pg-circuit, 15 is a registered ministry-specific command and data dictionary, 16 is an input buffer memory, 17 is a selection editing circuit, and 18 is an output Section f memory, 19 an audio reproduction circuit, and 20 a command target device. 2F Ko 1 figure

Claims (1)

[Scope of claims] Separated from Oshitsu's trial, and selected from a detailed list of the nine semi-patterns #* prepared in response to the Ministry of Foreign Affairs, U1 was then prepared to be accepted into the Ministry of Foreign Affairs. Hi as a signal,
Send back the ZengP or signal sound, and then send the Iih rice recognized using the station standard pattern dictionary tube for each V insertion force from the Ministry of Cough and Talk to the Ministry of Nuclear Affairs through the speech synthesis output in the form of I, and confirm. A voice input method that imitates % intoxication. (2) In the above feature, the repeating N-durable audio output circuit is also used as the output Ig1% for responding to continuous output of (IPM monitoring, needle play data, and stored data of the target equipment). tr+IIl search range fi141 voice input method.