JPS62293298A - Man-machine dialogue apparatus - 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 [Object of the Invention] (Industrial Application Field) The present invention relates to a man-machine dialogue device that performs a voice dialogue in response to a user's voice input.

(Prior Art) In recent years, dialogue devices that perform dialogue using voice recognition and voice responses have been devised. Examples of such dialogue devices include services such as credit card balance inquiries using bank telephones.

This dialogue device includes a voice recognition unit that recognizes a voice input signal from the user, a task processing S that performs processing according to the recognized voice (in this case, balance inquiry processing), and the answer and usage It consists of a voice response section that generates questions for the user, and a conversation control section that controls the flow of the conversation (determines which message should be output next to the user, etc.). By inputting words such as numbers by voice according to the voice output from the voice response section, the user can know the bank balance through direct speech. In such a dialogue device, the task processing unit is used as a database crosspiece, the voice recognition unit recognizes commands related to it, and the conversation control unit and voice response unit also respond accordingly. It is also possible to do this.

(Problem to be solved by the invention) It takes a lot of time and money to process like the data pace crossbars, and also for speech recognition, it is necessary to recognize continuous speech bits instead of using M recognition. In such a case, the recognition process itself requires a lot of time at the current level of technology. In other words, there is no problem if it does not take much time to receive the corresponding message from the system after the user inputs the voice, but if the processing takes a long time, the user may The problem is that you don't know what's going on, and it feels very unnecessary.

The purpose of the present invention is to eliminate the anxiety that users feel when the time between when a user performs voice input and when a message corresponding to 7K is obtained from the system is too long.
Supplying man-machine dialogue equipment.

[Structure of the Invention] (Means for Solving the Problems) In the present invention, after the voice recognition section inputs the voice from the user, the task processing section processes it, and the corresponding message is sent from the voice response section. Before starting output, the specific word is recognized in the Mayuki speech recognition unit, and whether the specific word or g;a
At the time the voice response unit generates a reply with predetermined content.

(Function) In the present invention, while performing processing corresponding to the content of input speech, recognition is performed only on specific words, and when such a specific word is recognized, a specific message is sent to the user. Can be provided.

(Example) Hereinafter, an example of the present invention will be described based on the drawings. FIG. 1 is a block diagram of the configuration of a man-machine dialogue device according to an embodiment of the present invention. Speech recognition section 1, conversation control section 2. Task processing unit 3. Voice response section 4. It consists of register 5. The state of the speech recognition unit consists of two modes. One is a mode in which g-knowledge is performed using candy bars related to normal tasks, and the other is a mode in which specific words are recognized. Hereinafter, these modes will be referred to as the j1 normal mode and the specific mode, respectively. This mode is set in the register and is connected by the conversation control P1 section 2.
s2. In the initial state, this mode is set to the normal mode. When the user inputs voice, the voice recognition sl
informs conversation control 14] section 2 via connection i 16;
The conversation control unit 2 sets a specific mode in the register. The speech recognition section 1 sends the recognition result of the user's input speech to the conversation control section 2 via the connection line 11. In response to this input, the conversation control section 2 issues a command to the task processing section 3 via the connection line 13 to start the corresponding process.

The task processing section 3 performs corresponding processing (for example, database cross-section) and sends the results obtained to the conversation control section 2 via a connection line 14.
When the conversation control unit 2 obtains the result, it sets the register 5 to the normal mode, and then outputs the result as a voice from the voice response unit 4. Conversation is carried out by repeating this process.

When the speech recognition unit 1 enters the specific mode, it recognizes only specific words (for example, "moshimoshi").

When this specific word is entered by the user, voice recognition #! The capital 1 informs the conversation control s2 of this via the connection line 16. At this time, the conversation control s2 activates the voice response unit 4 to generate a specific self-answer (for example, "Please wait a little longer").

As a result, between the time the user inputs voice input and the corresponding message is received, if the user releases a candy that says ``Moshi Moshi'', the system will generate an apology saying ``Please wait a little longer''. can.

In this embodiment, since the speech recognition unit l is an embodiment assuming recognition of candy, etc., the time involved in the recognition process can be considered to be relatively short, and immediately after the end of the voice input from the user. , it is assumed that recognition results can be obtained. For other embodiments in which the recognition process requires time, the recognition process may be performed in parallel, especially in the constant mode.

Alternatively, it is also possible to implement the present invention without performing such parallel processing by separately preparing a recognition unit that recognizes only specific words.

FIG. 2 illustrates a second embodiment of the invention. The specific word speech recognition unit 26 operates when the contents of the register 25 are in a specific mode. The voice recognition unit 21, unlike the voice-H unit 1 of the first embodiment, does not have two operation modes and always performs normal speech recognition. Other blocks, conversation fileJ N section 22, task processing section 23. Voice response section 24. The registers 25 are the same as those in the first embodiment. Note that various specific words to be recognized and contents to be responded to can be considered.

〔Effect of the invention〕

As described above, according to the present invention, if it takes a long time after the user inputs the sound P to obtain the corresponding message, if the user says 1, ``Hoshihoshi'', ``Please wait a little longer''. Since the system can respond with "Please, please," the user can fully utilize the system without feeling anxious.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of the structure of a man-machine dialogue apparatus according to an embodiment of the present invention, and FIG. 2 is a block diagram of the structure of a second embodiment of the present invention. l, 21...Voice recognition section, 2.22...Conversation control s
13.23...Task processing @-4-24...Sound P response unit, 5.25...Register, 26...Specific word speech recognition unit. Agent Patent Attorney Norishige Chika

Claims (1)

[Claims] A man-machine dialogue device that includes a means for recognizing voice sent from a user, a means for performing processing corresponding to the content of the input voice, and a voice response unit for outputting an answer obtained by the processing to the user. While processing is being performed in accordance with the content of input speech, the former speech recognition device or a separately provided speech recognition device is only recognizing specific words, and when the specific words are recognized, A man-machine dialogue device, characterized in that a message with specific predetermined content is output from the voice response section.