JP2002073080A - Voice interactive system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a audio response system improved in comfort to a user by reducing the data size of dialog script data. SOLUTION: An interactive processing part is provided with an assistant mechanism 7, and the assistant mechanism 7 is provided with a global transition condition table 8 and a means 9 for generating an assistant interactive mode. A plurality of conditions that are used in common under any interactive states are stored in the table 8. The means 9 provides an assistant interactive mode that helps the user when a usual basic interactive mode does not run smoothly.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates to a voice interaction system. In recent years, a voice interactive interface has begun to spread rapidly as one form of a so-called man-machine interface for exchanging information between a human and a machine. As a typical example, there is a voice interactive interface applied to an audio device or a navigation device, which is an on-vehicle information device, and the operability of the on-vehicle information device for a user driving a vehicle is dramatically improved. ing.

[0002] The present invention describes a voice interactive system for realizing the above voice interactive interface.

[0003]

2. Description of the Related Art FIG. 18 is a diagram showing a prototype of a general speech dialogue system. In this figure, a speech dialogue system 1
Are roughly divided into a speech recognition unit 2, a dialog processing unit 3, and a speech synthesis unit 4. Further, the dialogue processing unit 3 is a "semantic understanding system"
And “dialogue management system”. The former has an utterance database 5 containing utterance model data as a main component, and the latter has a dialogue database 6 containing dialogue script data as a main component. still,
There is an idea that the “semantic understanding system” has a configuration in which the voice recognition unit 2 has the function.

In order to easily understand the prototype of FIG. 18, a fast food store will be described as an example. The voice recognition unit 2 is provided to face a customer, while the voice synthesis unit 4 is used as a substitute for a clerk of the store. Function as Assuming that a visitor utters “two burgers,” the voice recognition unit 2
The voice is converted into text data. In this case, it can be converted into [Burger] and [Futatsu].

[0005] With the speech recognition result (character code) as an input, the dialogue processing unit 3 first refers to the utterance database 5 to give the meaning of the character code. in this case,
[Hamburger = order] and [Futatsu = order number] can be used. The dialogue processing unit 3 determines the flow of the dialogue with reference to the dialogue database 6 based on the meaning given above. In other words, it decides what kind of dialogue should be given to the visitor subsequently. In this case, for example, [Do you have it here? ] And [Drinks? 〕It turns out that. The dialogue determined here becomes a voice synthesized electronically by the voice synthesis unit 4 and is returned to the customer. Visitors answer this further, and the dialogue changes one after another.

[0006]

The above-mentioned general voice dialogue system has been described using a fast food restaurant as an example. However, for example, in the above-described navigation device, the content of the dialogue is quite complicated and various, and the flow of the dialogue is complicated. Also transitions to several levels. For this reason, further improvement is required for the voice dialogue system. Although the points to be improved are various, the present invention focuses on the interactive script data as points to be improved.

[0007] First, conventionally, in the transitioning dialog flow, dialog script data is prepared individually for all possible patterns. Second, when a normal recognition result cannot be obtained from the voice recognition unit 2,
Standard conversation script data alone cannot continue the normal conversation flow any further.

As for the first point, there is a problem that the data size of the interactive script data becomes extremely large and the memory capacity of the interactive database 6 becomes enormous (first problem). As for the second point, since the dialogue does not flow smoothly, there is a problem that the user is uncomfortable (second problem).

SUMMARY OF THE INVENTION Accordingly, it is an object of the present invention to realize a voice interactive system which reduces the data size in an interactive database and can provide a comfortable dialog to a user. .

[0010]

FIG. 1 is a diagram showing the concept of the present invention. As shown in the drawing, a feature of the present invention is that an assist mechanism 7 is provided in a dialog processing unit (corresponding to 3 in FIG. 18). The assist mechanism 7 includes a global transition condition table 8 and an assist dialog mode generation means 9.

The global transition condition table 8 stores in advance a plurality of global transition conditions that are commonly used under any conversational state in the utterance state of the user. The dialogue processing unit accesses the global transition condition table 8 and generates the dialogue content when the content of the utterance of the user in the utterance state corresponds to the global transition condition.

This eliminates the need to describe all dialog transitions in the dialog script data, and can solve the first problem described above. On the other hand, the assist interaction mode generation means 9 enables the interaction processing unit to operate not only in the basic interaction mode but also in the assist interaction mode. Here, the basic interaction mode is a mode for forming a flow of a normal series of interaction for collecting various information requested by an application program such as a navigation device,
The assist dialogue mode is a mode formed at the time of an abnormality in which the flow of a series of dialogues cannot be maintained.

Thus, the user can smoothly progress the flow of a series of dialogues and the comfort is improved, so that the second problem described above can be solved.

[0014]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 2 is a diagram showing an example of the overall configuration of a voice interaction system according to the present invention. As shown in this figure,
The voice interaction system 1 according to the present invention basically recognizes a voice input unit 11 for inputting a voice uttered by a user in a utterance state via a microphone, for example, and recognizes a voice from the voice input unit 11. Based on the recognition result by the voice recognition unit 12, a dialog content to be returned to the user is generated with reference to the dialog script data stored in the dialog database 14 in advance. A dialogue processing unit 13 and a voice output unit 16 for returning a synthesized voice to the user based on the content of the dialogue are provided. Each time the user utters, a series of dialogue states defined by the dialogue script data are successively changed. This is a voice interactive system that collects various information requested by an application program from a user in an interactive manner.

The first feature of the present invention is to further provide a global transition condition table 15 (corresponding to 8 in FIG. 1) which is referred to by the interaction processing unit 13.
In the global transition condition table 15, a plurality of global transition conditions that are commonly used under any conversation state in the utterance state are stored in advance. Here, when the content of the utterance in the utterance state corresponds to the global transition condition, the dialog processing unit 13 accesses the global transition condition table 15 and generates the content of the dialog.

Before describing the specific contents of the global transition condition table 15, the components of FIG. 2 will be described in more detail. The speech input from the speech input unit 11 composed of a microphone is cut out into each word by the continuous word recognition engine 17 constituting the speech recognition unit 12 (2 in FIG. 18) and converted into text. This text is input to an utterance adaptation unit 18 also constituting the speech recognition unit 12. The unit 18 refers to the utterance model data (5 in FIG. 18), understands the meaning of the input text (word), and outputs the recognition result to the interactive processing unit 13.
(Corresponding to 3 in FIG. 18). For example, in a navigation device, if the recognized text (voice) is “want to go”, the recognition result is “destination”.

The dialogue processing unit 13 that has obtained the recognition result transmits the dialogue script data (FIG. 18) in the dialogue database 14.
Based on 6), the flow of the dialog is determined and assembled. For example, for the above “Destination”, “Amusement park?”
→ "△△ prefecture?" → "○ land?" Since both proceed in the form of dialogue with the user, the next dialogue is voice-synthesized by the voice synthesis unit 19 (4 in FIG. 18) constituting the voice output unit 16 and returned to the user from the speaker.

The result obtained in the flow of a series of dialogues, for example, “destination / amusement park / △△ prefecture / ○ land” is given to the application program for operating the navigation device as a dialogue result. Next, the global transition condition table 15 will be described. Global transition condition table 1
5 specifies any one of a plurality of global transition files that store individual conversation contents corresponding to each of a plurality of predetermined global transition conditions.

Some examples of the plurality of global transition conditions are as follows. i) No recognition ii) Recognition error iii) Return (want to return) iv) Cancel (cancel) v) Don't know vi) Mistake i) to ii) above are global transition conditions issued by the voice recognition unit 12 by itself. , Iii) to vi) are global transition conditions issued from the user side. Each of these cases is a common event in any interaction pattern. For this reason, it is named "global".

Conventionally, the above-mentioned global transition condition is provided for all conversation patterns, so that the data size of the conversation database 15 becomes extremely large. FIG. 3 is a diagram schematically showing a global transition file. Global transition files G-1, G-2 in this figure
To G-6 are the above global transition conditions i), ii) to v
i) respectively. However, G-3 to G-5
Are not shown.

Note that these files G1 to G-6 are
It may be additionally formed in the interactive database 14 of FIG. 2, or may exist independently as files G-1 to G-6. In any case, the interaction processing unit 13 reads the contents of the file via the global transition condition table 15. FIG. 4 is a diagram schematically showing the contents of the global transition condition table 15.

If the result of recognition by the voice recognition unit 12 is "not recognized", it specifies that the file G-1 should be accessed. “Not recognizing” means that the user does not utter anything while speaking. If the recognition result by the voice recognition unit 12 is “recognition error”, it specifies that the file G-2 should be accessed. The “recognition error” means that the user does not speak, but the surrounding noise is too loud, the voice is too loud, or the voice is too loud, so that the voice recognizing unit 12 cannot recognize correctly.

When the user utters "return", the program jumps to the file G-3 and enters a dialog sequence for returning to the previous dialog state. When the user utters “stop”, the process jumps to the file G-4 and enters a dialog sequence for ending the dialog. FIG. 5 is a diagram schematically illustrating an example of the data structure of the global transition file. However, only the files G-1 and G-2 are shown.

As shown in the figure, the file G-1 selects "S-5" of the dialog sequence of the dialog script data and "2" of the node ID in the dialog sequence "S-5". Is instructed to be selected. The same applies to the file G-2,
A conversation ID “S-21” and a node ID “4” therein are described.

That is, the global transition file G-
, G-2... Include dialog ID information specifying a transition destination dialog ID and node ID information specifying a transition destination node ID in a series of dialog sequences developed by the specified transition destination dialog ID. Store. FIG. 6 is a diagram illustrating an example of the interaction sequence. For example, the interaction sequence S-1
Is a sequence of a normal basic conversation mode, and a destination is set at the node 1 by, for example, utterance of a user who wants to go.

The node 2 asks the user "Please go to the destination". At node 3, the user says "Hyogo prefecture". At node 4,
Ask the user "Please strip off the municipal name."

Hereafter, the dialogue continues. Originally, the flow of such a series of dialogues should proceed. If the recognition result from the speech recognition unit 12 in the node 2 is “not recognized”, the dialogue processing unit 13 15 is accessed. In this case, as shown in FIG. 5, a dialogue sequence corresponding to "not recognized" of the dialogue number G-1 is started, and re-instruction is drawn from the user according to predetermined guidance. Thereafter, the transition destination conversation ID “S-5” written in the global transition file G-1
And the transition destination node ID "2", the double circle "2" in FIG.
Transition to.

FIG. 7 is a diagram schematically showing a modification of the data structure of FIG. In the conversation number G-3, the previous conversation ID
The transition to the immediately preceding u (the utterance node from the dialogue processing unit 13 before moving) in (the dialogue ID before moving) is made. At the conversation number G-4, the transition is made to the immediately preceding u (the utterance node of the user before the movement) in the immediately preceding conversation ID (the conversation ID before the movement).

That is, based on the modified example of FIG. 7, the above-described dialog ID information and the above-described node ID information respectively represent various transition destinations in a functional expression (PREV-previo).
us). By expressing the functions in this manner, it is necessary to specifically specify the ID of the interaction sequence in each case, and the data size of the files (G-1 to G-6) can be further reduced.

Next, the assist dialog mode generation means 9 in the assist mechanism 7 shown in FIG. 1 will be described. The dialogue processing unit 13 which functions the assist dialogue mode generation means 9
Can operate in the assist interactive mode in addition to the basic interactive mode. Here, the basic interaction mode is a mode for forming a normal series of interaction flows for collecting various types of information requested by the application program (see the upper part of FIG. 2), and the assist interaction mode is for forming the series of interaction flows. This mode is formed when an abnormality cannot be maintained.

FIG. 8 is a flowchart showing a first operation mode by the assist dialog mode generation means 9. In the first operation mode, the dialogue processing unit 13 shifts to the assisted dialogue mode and operates when the speech recognition unit 12 continuously gives a recognition result indicating “not recognized”.

Referring to FIG. 8, step S11: The dialog processing unit 13 determines whether or not the recognition result "not recognized" received from the speech recognition unit 12 is the first time. Step S12: If the determination result is No, that is, the recognition result of “not recognized” is given twice or more,
When continuously received, the interaction processing unit 13 shifts to the assist interaction mode under the first operation mode.

Step S13: When the result of the determination is Yes, that is, when the recognition result of "not recognizing" is received for the first time, the process returns to the previous conversation ID and node ID. The above-described assist interactive mode will be specifically described. Now, suppose that the question in the basic dialogue mode is "Please tell me the name of the prefecture", and that the result of recognition by the voice recognition unit 12 for the response from the user to this was "No recognition" continuously. Then, the mode shifts to the assist interactive mode. In this assist interaction mode, the above-mentioned question is changed to another way. For example, for the above example, change it to another phrase such as "Please talk like Osaka or Hyogo prefectures".

As a result, the user can continue speaking and a comfortable dialogue flows smoothly. FIG. 9 is a flowchart showing a second operation mode by the assist interactive mode generation means 9. In the second operation mode, the dialog processing unit 13 operates in the stop dialog mode for stopping the voice input when the recognition result indicating “there is a recognition error” is continuously given from the voice recognition unit 12. .

Referring to FIG. 9, step S21: The dialog processing unit 13 determines whether the recognition result "recognition error" received from the speech recognition unit 12 is the first recognition result. Step S22: If the determination result is No, that is, if the recognition result of “recognition error” is received twice or more consecutively, the interaction processing unit 13
To the suspension interactive mode under the operation mode of (1).

Step S23: When the result of the determination is Yes, that is, when a recognition result of "recognition error" is received for the first time, the process returns to the previous conversation ID and node ID. The stop interactive mode will be specifically described.
When "recognition errors" continue in response to the inquiry in the basic dialogue mode, the user can be assisted in various stop dialogues in the stop dialogue mode. A few examples are as follows.

"Your voice is difficult to hear, so stop the voice input (or stop it?)""Your voice cannot be recognized correctly""Please make it a little bigger and clearer" .

Thus, the user can know the reason why the conversation does not flow smoothly, and is released from the conventional uncomfortable waiting time. FIG. 10 is a flowchart showing the third and fourth operation modes by the assist dialog mode generation means 9. In the third operation mode, the dialogue processing unit 13 executes a timeout process when a recognition result is not given even after a predetermined period of time has elapsed from the voice recognition unit 12.

In this case, if the mode for a certain period of time is the basic interactive mode, the assist interactive mode is formed by executing the above timeout process. Further, the fourth operation mode appears when the mode for the certain time is the assist interactive mode. At this time, the execution of the above-mentioned timeout processing shifts to a paused dialogue for temporarily stopping the dialogue.

Referring to FIG. 10, step S31: dialogue processing section 13 determines whether or not a recognition result has been received from speech recognition section 12. If it has been received (Yes), the process proceeds to the next conversation (END). Step S32: If the above-mentioned reception has not been performed (No), the dialogue processing unit 13 executes the above-mentioned timeout processing and determines whether or not the above-mentioned certain time has elapsed. If it has not elapsed (No), steps S31 and S32 are repeated.

Step S33: If the above-mentioned fixed time has elapsed (Yes), the dialog processing unit 13 determines whether or not the (current) mode in the certain time is the basic dialog mode. Step S34: If the result of the determination indicates that the current mode is the basic interactive mode (Yes), the process shifts to the assist interactive mode. Step S35: If the result of the determination indicates that the mode is not the basic dialogue mode (No), the process shifts to the paused dialogue.

Thus, if the above-mentioned step S34 is entered,
The user can be guided to a dialog to assist the user. If the process goes to step S35, the dialogue processing unit 13
Returns, for example, a message "Pause the dialog here" to the user, and the user can confirm the situation in which the user is located. FIG. 11 is a flowchart showing fifth and sixth operation modes by the assist dialog mode generation means 9.

In the fifth mode of operation, the fourth mode
Under the paused dialogue for temporarily stopping the dialogue in the operation mode, when the user turns on the dialogue resume switch (SW in FIG. 1), the timeout processing is executed. Then, the process shifts to the central dialogue for stopping the voice input. In the sixth operation mode,
When the above-described predetermined time has elapsed again during the operation in the assist interaction mode, the interaction processing unit 13 shifts to another assist interaction mode having a more detailed interaction content different from the assist interaction mode.

Referring to FIG. 11, step S41: dialogue processing section 13 determines whether or not a recognition result has been received from speech recognition section 12. If it has been received (Yes), the process proceeds to the next conversation (END). Step S42: If the above-mentioned reception is not performed (No), the interaction processing unit 13 executes the above-mentioned timeout processing and determines whether or not the above-mentioned certain time has elapsed. If it has not elapsed (No), steps S41 and S42 are repeated.

Step S43: If the above-mentioned fixed time has elapsed (Yes), the dialog processing unit 13 determines whether or not the mode (current mode) at that fixed time is a paused dialog. Step S44: If the result of the determination indicates that the dialogue is a paused dialogue (Yes), the process shifts to the above-mentioned suspended dialogue.

Step S45: If the result of the determination indicates that the conversation is not a paused conversation (No), the dialog processing unit 13 determines whether or not the current conversation state is in the assist dialog mode. If No, go to a) of FIG. Step S46: If the result of the above determination is Yes,
A transition is made to another assist interaction mode having a different and more detailed interaction content as described above.

Thus, the user can receive a message whose expression is further changed. FIG. 12 is a flowchart showing seventh and eighth operation modes by the assist dialog mode generation means 9. In the seventh operation mode, the dialog processing unit 13 shifts to the assist dialog mode when a recognition result indicating a specific voice input is given from the voice recognition unit 12 during the utterance state of the user.

In the eighth mode of operation, the dialogue processing unit 13 sets a specific switch (FIG. 1) while the user is speaking.
When the switch SW2) is detected to be turned on, the mode shifts to the assist interactive mode. Referring to FIG. 12, step S51: the dialog processing unit 13 determines whether or not the user is in a dialog state in which a voice input is received from the user. If the result of the determination is No, the process ends (END).

Step S52: The above judgment result is Yes
Then, it is determined whether or not the voice recognition unit 12 has received a voice input recognition result indicating transition to the assist dialog mode. Step S53: If the result of the above determination is Yes, shift to the assist interactive mode.

Step S54: If the result of the above determination is No
Then, it is determined whether the specific switch SW2 is turned on by the user. If this determination result is No, the process ends (END), but if Yes, the process shifts to the assist interactive mode. In this way, the user can obtain an assist message having a much deeper content from the audio output unit 16.

The specific switch SW2 may be a mechanical switch or a touch panel type switch, or a GUI (Graphical Use).
rInterface). If it is troublesome for the user to give an instruction by voice, such a switch operation can be used. FIG. 13 is a flowchart showing a ninth operation mode by the assist interaction mode generation means 9, and FIG. 14 is a diagram schematically showing the ninth operation mode.

Under the ninth mode of operation, the dialogue processing unit 13 stores the use histories of the basic dialogue mode and the assistive dialogue mode in, for example, a memory (not shown), When the usage frequency exceeds the usage frequency of the basic interactive mode, the operation is performed so that the assist interactive mode and the basic interactive mode are switched.
First, referring to FIG. 14, it is assumed that a certain basic dialog has been written in the dialog script data so as to proceed in the order of dialog nodes 1 → 2 → 4. Then, it is assumed that the assist dialog corresponding thereto has been similarly written so as to proceed as dialog nodes 1 → 3 → 5.

Here, it is assumed that when the use history of each dialog node is taken, it is found that the use frequency of the dialog nodes 1 → 3 → 5 is higher than the use frequency of the dialog nodes 1 → 2 → 4. Then, based on the above fact, the dialogue processing unit 13 starts the dialogue node 3 when it comes to the dialogue node 1.
→ Set the sequence to transition to 5.

In this way, a dialog sequence closer to the user's preference can be provided to the user from the beginning, and the comfort is improved. Next, referring to FIG. 13, step S61: It is determined whether or not the number of nodes 3 is greater than the number of nodes 2 with respect to the number of dialog nodes in FIG.

Step S62: The result of the above judgment is Ye
If s, the dialog nodes 2 and 3 are exchanged. Step S63: If the result of the above determination is No, the current state is maintained. FIG. 15 shows the assist interactive mode generating means 9.
For explaining the tenth operation mode according to (1)
FIG. 16 is the same figure (No. 2), and FIG. 17 is the same figure (No. 3).

In the tenth mode of operation, when the basic interactive mode and the assist interactive mode are each composed of a series of a plurality of interactive nodes having a hierarchical structure, the basic interactive mode and the assist interactive mode For at least one mode, the usage history of each dialog node is stored in, for example, a memory (not shown), and the hierarchy is sequentially switched from the node of the higher usage frequency to the node of the lower usage frequency.

Referring first to FIG. 15, in the example of FIG.
It shows the flow of the dialog that transits to the dialog node 2,
The node 1 → 2 → 8 belongs to the basic conversation mode and the node 3
→ 4 → 5 → 7 belongs to the assist interactive mode. Here, it is assumed that the use history of each dialog node that has shifted to the dialog node 2 has a result as shown in FIG. 16, for example. Then, finally, the number of times of passing through the dialog node 5 is four times, similarly, the dialog node 4 is three times, and the dialog node 3 is two times, and the dialog nodes are 5 → 4 → It becomes 3.

Then, when the dialog nodes are rearranged based on the above result, the result is as shown in FIG. In this way, the flow automatically shifts to the most natural flow of conversation for the user, and the comfort is improved.

[0059]

As described above, according to the present invention, firstly, the data size of the dialog script data stored in the dialog database 5 can be significantly reduced as compared with the related art. Second, it is possible to provide a more comfortable conversation environment for the user than before.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the concept of the present invention.

FIG. 2 is a diagram showing an example of the overall configuration of a voice interaction system according to the present invention.

FIG. 3 is a diagram schematically showing a global transition file.

FIG. 4 is a diagram schematically showing the contents of a global transition condition table 15;

FIG. 5 is a diagram schematically illustrating an example of a data structure of a global transition file.

FIG. 6 is a diagram illustrating an example of a dialog sequence.

FIG. 7 is a diagram schematically showing a modification of the data structure of FIG. 5;

FIG. 8 is a flowchart showing a first operation mode of the assist interactive mode generation means 9;

FIG. 9 is a flowchart showing a second operation mode by the assist interactive mode generation means 9;

FIG. 10 is a flowchart showing third and fourth operation modes by the assist interactive mode generation means 9;

FIG. 11 is a flowchart showing fifth and sixth operation modes by the assist interactive mode generation means 9;

FIG. 12 is a flowchart showing seventh and eighth operation modes by the assist interactive mode generation means 9;

FIG. 13 is a flowchart showing a ninth operation mode of the assist interactive mode generation means 9;

FIG. 14 is a diagram schematically showing a ninth operation mode.

FIG. 15 shows a tenth dialogue generated by the assist interactive mode generation means 9;
FIG. 3 is a diagram (No. 1) for describing the operation mode of FIG.

FIG. 16 shows a tenth dialogue generated by the assist dialogue mode generation means 9;
FIG. 10 is a diagram (part 2) for explaining the operation mode of FIG.

FIG. 17 shows a tenth dialogue generated by the assist dialogue mode generation means 9;
FIG. 10 is a diagram (No. 3) for describing the operation mode of FIG.

FIG. 18 is a diagram showing a prototype of a general spoken dialogue system.

[Explanation of symbols]

 7 Assist function 8 Global transition condition table 9 Assist dialog mode generation means 11 Voice input unit 12 Voice recognition unit 13 Dialog processing unit 14 Dialog database 15 Global transition condition table 16 Voice output unit 17 Continuous Word recognition engine 18 Speech adaptation unit 19 Speech synthesis unit SW1 Dialogue restart switch SW2 Specific switch

──────────────────────────────────────────────────続 き Continuation of the front page (51) Int.Cl. ⁷ Identification symbol FI theme coat ゛ (reference) G10L 15/28 G10L 3/00 551Q // G01C 21/00 F term (reference) 2F029 AA02 AB13 AC18 5D015 KK02 KK04 5D045 AB21 AB30 5E501 AA20 AA22 AA23 AC03 BA05 BA12 CA08 CB15 DA11 EA21

Claims (15)

[Claims] 1. A voice input unit for inputting voice uttered by a user in a utterance state, a voice recognition unit for recognizing the voice from the voice input unit and analyzing its meaning, and a recognition by the voice recognition unit. A dialogue processing unit that generates dialogue contents to be returned to the user while referring to the dialogue script data stored in the dialogue database in advance based on the result; and returns a synthesized voice to the user based on the dialogue contents. A voice output unit, wherein each time the user utters a voice, a series of dialogue states defined by the dialogue script data are successively transited, and various information requested by the application program is collected in an interactive manner from the user. In the voice interaction system, a global transition condition table referred to by the interaction processing unit is further provided. In the global transition condition table, in the utterance state, a plurality of global transition conditions that are commonly used under any of the dialogue states are stored in advance, and the dialogue processing unit includes a content of the utterance in the utterance state. Wherein the dialogue content is generated by accessing the global transition condition table when the global transition condition is satisfied. 2. The global transition condition table according to claim 1, wherein the global transition condition table specifies a plurality of global transition files that store individual conversation contents corresponding to each of a plurality of predetermined global transition conditions. Spoken dialogue system. 3. Each of the plurality of global transition files includes dialog ID information for specifying a transition destination dialog ID, and a transition destination node ID in a series of dialog sequences developed by the specified transition destination dialog ID. 3. The voice interaction system according to claim 1, wherein the node ID information to be specified is stored. 4. The dialog ID information and the node ID
4. The voice interaction system according to claim 3, wherein the information is described by codes representing various transition destinations. 5. A voice input unit for inputting a voice uttered by a user in an utterance state, a voice recognition unit for recognizing the voice from the voice input unit and analyzing the meaning, and a recognition by the voice recognition unit. A dialogue processing unit that generates dialogue contents to be returned to the user while referring to the dialogue script data stored in the dialogue database in advance based on the result, and returns a synthesized voice to the user based on the dialogue contents A voice output unit, and each time the user utters a voice, a series of dialogue states defined by the dialogue script data are sequentially transited, and various information requested by the application program is collected in an interactive manner from the user. In the voice interaction system, the interaction processing unit operates in an assist interaction mode in addition to the basic interaction mode. The present interactive mode is a mode for forming a normal series of dialog flows for collecting various information requested by the application program, and the assist interactive mode is formed when an abnormality cannot be maintained for the series of dialog flows. A speech dialogue system characterized by being in a mode. 6. The dialogue processing unit operates when the speech recognition unit continuously receives a recognition result indicating “not recognized” from the voice recognition unit and shifts to the assisted dialogue mode. Item 6. The voice interaction system according to Item 5. 7. The interactive processing unit operates in a stop interactive mode for stopping a speech input when a recognition result indicating “there is a recognition error” is continuously given from the voice recognition unit. The voice interaction system according to claim 5, wherein 8. The dialogue processing unit executes a timeout process when the recognition result is not given even after a predetermined period of time has elapsed from the voice recognition unit, and the mode in the predetermined period is 6. The voice interaction system according to claim 5, wherein when in the basic interaction mode, the assist interaction mode is formed by executing the timeout process. 9. The dialogue processing unit executes a time-out process when the recognition result is not given even after a predetermined period of time has elapsed from the voice recognition unit, and the mode in the predetermined period is set to the mode. The voice interaction system according to claim 5, wherein, when the assist interaction mode is set, the execution of the timeout process causes a transition to a suspension interaction for temporarily suspending the interaction. 10. When the timeout process is executed after turning on a dialogue resume switch under the paused dialogue for temporarily suspending the interaction, the voice input is stopped by executing the timeout process. The voice dialogue system according to claim 9, wherein the process shifts to a stop dialogue to perform the dialogue. 11. The method according to claim 1, wherein the dialogue processing section shifts to another assistive dialogue mode having a dialogue content different from that of the assistive dialogue mode when the predetermined time has elapsed again while operating in the assistive dialogue mode. Spoken dialogue system. 12. The method according to claim 1, wherein the dialogue processing unit shifts to an assistive dialogue mode when a recognition result indicating a specific voice input is given from the voice recognition unit while the user is speaking in the assistive dialogue mode. The voice interaction system according to claim 5, wherein 13. The voice dialogue according to claim 5, wherein when the dialogue processing unit detects that a specific switch is turned on during a utterance state of the user, the dialogue processing unit shifts to an assistive dialogue mode. system. 14. The usage history of each of the basic interaction mode and the assist interaction mode is stored, and when the usage frequency of the assist interaction mode exceeds the usage frequency of the basic interaction mode, the assist interaction mode and the basic interaction mode are stored. The voice interaction system according to claim 5, wherein the conversation mode is switched. 15. When the basic interaction mode and the assist interaction mode are respectively composed of a series of a plurality of interaction nodes having a hierarchical structure, each of the basic interaction mode and the assist interaction mode has at least one of: The voice dialogue system according to claim 5, wherein the use history of the dialogue nodes is stored, and the hierarchy is sequentially switched from a node having a higher frequency of use to a node having a lower frequency.