JP2001109493A - Voice interactive device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that a conventional voice interactive devices where a keyword mutual relation extending over plural utterance is not considered, a user is recognized at every utterance for improving a interactive success rate, multiple recognition communication are executed and the convenience of the user and the naturalness of communication are damaged. SOLUTION: The interactive device is provided with a interactive procedure storage part storing the vocabulary of a recognition object in respective interactive states, a system response, a system response assumption response and a transition destination interactive state corresponding to the response, a voice recognition part recognizing voice with the vocabulary of the recognition object in the respective interactive states of the interactive procedure storage part and outputting plural results, a transition destination interactive state deciding operation deciding part deciding a transition destination interactive state according to the recognition result of the voice recognition part and the content of the interactive procedure storage part, deciding one state when the hypothesis of the transition destination interactive state satisfies a prescribed condition, holding decision when it does not satisfy the prescribed condition and outputting the hypothesis of the shift destination interactive state, and an interactive operation execution part outputting the system response recognizing the recognition result of the hypothesis of the transition destination interactive state when hypothesis is decided and outputting the system response of the hypothesis of the transition destination interactive state when decision is held.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech dialogue processing device used for a man-machine interface using a natural language.

[0002]

2. Description of the Related Art Speech dialogue devices that obtain information required by a user through voice dialogue have become increasingly important. In such a spoken dialogue apparatus, it is important to recognize keywords included in a user's input with high precision and to perform dialogue control so that the user can obtain necessary information efficiently.

Conventionally, there has been proposed a method of utilizing the relevance between a plurality of keywords included in input speech in order to recognize a keyword with a high accuracy rate. For example, FIG. 14 shows a recognition candidate extracting device disclosed in Japanese Patent Laid-Open No. 7-92994.
In the conventional recognition candidate extraction device configured as described above, the speech recognition device recognizes a plurality of keywords from a continuous speech including a plurality of keywords having relevance, and ranks each keyword in descending order of recognition likelihood. Output multiple recognition results.

[0004] A candidate extraction processing device extracts only a set of keywords that match the combination information from the recognition result output from the speech recognition device, using combination information between predetermined keywords, as a recognition candidate. By outputting and rejecting a combination of irrelevant keywords, the accuracy rate of recognition candidates is improved.

[0005] Further, the confirmation processing device confirms the recognition candidate output from the candidate extraction processing device by repeating the confirmation to the user, and if it is determined that the repetition recognition candidate is not correct, a plurality of keywords are determined. After the first keyword is recognized by the speech recognition device and the result of the recognition is determined by the determination processing device, only the keywords that can be combined with the determined keyword are extracted as recognition candidates in the second and subsequent keywords. Therefore, the accuracy rate of the recognition candidate is improved.

[0006]

However, in the conventional candidate extraction processing apparatus as described above, the relationship between keywords over a plurality of utterances is not taken into account, and it is obtained by several exchanges with the user. In order to improve the recognition rate of all input items and the success rate of dialogue, it is necessary to confirm with the user for each utterance and proceed deterministically. There was a problem of spoiling the naturalness of dialogue.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems, and has as its object to obtain a voice interactive device that can improve the recognition rate without performing a confirmation process for each utterance. is there.

[0008]

A speech dialogue apparatus according to the present invention includes a speech recognition section, a dialogue procedure storage section, a transition destination dialogue state determination action determination section, and a dialogue action execution section. This is a spoken dialogue device that obtains information required by a user through a dialogue. The dialogue procedure storage unit stores the vocabulary to be recognized in each dialogue state, the system response, the answer assumed for the system response, and the answer. The speech recognition unit defines and stores the transition destination conversation state, and performs speech recognition on the input speech using the recognition target vocabulary corresponding to each conversation state stored in the conversation procedure storage unit. Output recognition result,
The destination dialog state determination operation determining unit determines the destination dialog state based on the recognition result from the speech recognition unit and the contents of the dialog procedure storage unit, and when the hypothesis of the destination dialog state satisfies a predetermined condition. Is determined as one, and if the predetermined condition is not satisfied, the determination is suspended and the transition destination dialog state hypothesis is output. It outputs a system response for confirming the recognition result of the transition destination dialog state hypothesis from the decision operation determining unit, and outputs a system response of the transition destination dialog state hypothesis when suspending the decision.

Further, the speech dialogue apparatus according to the present invention further includes a dialogue state transition storage unit, a transitional destination dialogue state determination unit, and a provisional transitional destination dialogue state determination unit. , Determine whether the hypothesis of the transition destination dialog state determined from the recognition result from the speech recognition unit and the contents of the dialog state transition storage unit or the dialog procedure storage unit is to be determined as one or suspend the determination. Outputs the state hypothesis, and the transition destination dialog state determination unit inputs the transition destination dialog state hypothesis when the transition destination dialog state hypothesis from the transition destination dialog state determination operation determination unit is determined as one, and provides the user with the By confirming the recognition result, the transition destination dialog state is determined and output, and the stored transition destination dialog state hypothesis is rewritten in the dialog state transition storage unit. Transition destination dialog state determination operation When the transition destination dialog state hypothesis from the fixed part is suspended, the transition destination dialog state hypothesis is input, the provisional transition destination dialog state is determined and output, and the transition state dialog state is stored in the dialog state transition storage unit. The hypothesis is rewritten, and the dialog state transition storage unit stores the transition state transition history from the dialog start time and the transition destination dialog state hypothesis from the transition destination dialog state determination unit or the provisional transition destination dialog state determination unit, and performs the dialog operation. The execution unit receives the transition destination dialog state from the transition destination dialog state determination unit or the provisional transition destination dialog state determination unit, outputs a system response defined in the transition destination dialog state, and outputs the transition destination conversation state. The recognition target vocabulary specified in the above is output to the speech recognition unit, and the speech recognition unit performs speech recognition on the input speech using the recognition target vocabulary input from the dialogue execution unit, and obtains a plurality of recognition results. To It is intended to force.

[0010] Further, in the voice interaction apparatus according to the present invention, the voice recognition unit is configured to output a plurality of recognition results and a score of the recognition results, and the transition destination dialog state determination operation determination unit includes the voice recognition unit. It is determined whether or not to execute the finalizing operation according to the score of the recognition result input from the CPU.

Further, in the speech dialogue apparatus according to the present invention, it is necessary to perform a definite operation in advance in each dialogue state stored in the dialogue procedure storage unit in order to make a state transition from another dialogue state to the dialogue state. The destination dialog state determination operation determination unit describes whether or not there is, and the hypothesis of the destination dialog state determined from the recognition result input from the speech recognition unit, the contents of the dialog state transition storage unit, and the dialog procedure. However, when it is necessary to perform the determining operation in advance, it is determined that the determining operation is performed.

Further, in the voice interactive device according to the present invention, the transition destination interactive state determining operation determining unit determines the recognition result from the voice recognizing unit even if all the input items from the user are not input. When the item value for an uninput item is uniquely determined, the determination operation is determined to be performed.

Further, in the voice interaction apparatus according to the present invention, the transition destination dialog state determination operation determining unit determines whether to perform the determination operation according to a system response defined in the transition destination dialog state hypothesis. is there.

Further, in the voice interaction apparatus according to the present invention, the transition destination dialog state determination operation determination unit determines that the determination operation is performed when there is no common system response in the transition destination dialog state hypothesis, and If there is a common system response in the state hypothesis, only the transition destination dialog state hypothesis having a common system utterance is output as the transition destination dialog state hypothesis.

Further, in the speech dialogue apparatus according to the present invention, each dialogue state stored in the dialogue procedure storage unit can describe a plurality of system responses, and the dialogue operation execution unit transmits the dialogue state from the temporary transition destination dialogue state determination unit. When the destination dialog state is input, of the system responses specified in the input destination dialog state, the common system responses are the same as the system response specified in the destination dialog state hypothesis stored in the dialog state transition storage unit. It is to output things.

Also, in the voice interaction device according to the present invention, the transition destination dialog state determination operation determining unit determines that the total vocabulary of all the recognition target words of the transition destination dialog state hypothesis is larger than a predetermined reference. It is determined that the fixing operation is performed.

Further, in the speech dialogue apparatus according to the present invention, the transition destination dialog state determining operation determining unit refers to the dialog state transition storage unit to generate a transition sequence from the determined dialog state to the transition destination dialog state hypothesis. When the length is equal to or larger than a predetermined reference value, it is determined that the fixing operation is performed.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1. FIG. 1 is a block diagram showing Embodiment 1 of a voice interaction apparatus according to the present invention. 1 performs speech recognition on the input speech using the recognition target vocabulary input from the interactive operation execution unit described later,
A speech recognition unit that outputs a plurality of recognition results and scores of the recognition results, 2 is a dialogue procedure storage unit that specifies the vocabulary to be recognized in each dialogue state, a system response, and a transition destination dialogue state according to the speech recognition result. A conversation state transition storage unit for storing a conversation state transition history from the start of the conversation and a transition destination conversation state hypothesis; and a recognition result from the speech recognition unit as an input, and the recognition result and the conversation state transition storage unit. Is determined whether the hypothesis of the destination dialog state determined from the contents of the above and the dialog procedure is determined as one, and if it is determined, the hypothesis of the transition destination dialog state is output to a destination dialog state determination unit described later. The transition destination dialog state determining operation determining unit outputs the transition destination dialog state hypothesis to the temporary transition destination dialog state determining unit when the determination is suspended.

Reference numeral 5 designates a transition destination dialog state hypothesis from the transition destination dialog state determination operation determination section as input, confirms a recognition result to a user to determine and output a transition destination dialog state, and outputs the dialog state. A transition destination dialog state determination unit that deletes all stored transition destination dialog state hypotheses from the state transition storage unit and writes the determined transition destination dialog state.

[0020] A transition destination dialog state hypothesis from the definite operation determination unit is input, a provisional transition destination dialog state is determined and output based on the score of the recognition result, and the dialog state transition storage unit is determined. A transition destination dialog state hypothesis, which writes a transition destination dialog state hypothesis to the transition destination dialog state determination unit or the provisional transition destination dialog state determination unit. A system response defined in the state is output, and a recognition target vocabulary defined in the transition destination dialog state and a recognition target vocabulary defined in the transition destination dialog state hypothesis stored in the dialog state transition storage unit are output from the system. A dialogue execution unit that outputs to the voice recognition unit.

Hereinafter, a specific description will be given of the case where the present invention is used as a telephone number guidance voice interactive device.
A telephone number guidance voice interactive device is a device in which a user interacts with the device by voice, and inputs item values such as a prefecture name, a municipal name, a business type, and a target name necessary for telephone number guidance. The telephone number is searched based on the input item values, and the user is guided to the telephone number.

FIG. 2 shows an example of the dialogue state stored in the dialogue procedure storage unit. For example, in the dialogue state S1, R1 "Please name a prefecture" is defined as a system response, and a prefecture name is defined as a recognition target vocabulary V1. Also,
S2 as the transition destination dialog state when the recognition result is "Hokkaido"
Is stipulated. Hereinafter, an example will be described in which the number N of recognition results output by the voice recognition unit is 5, the threshold for the score of the first recognition result used by the transition destination dialog state determination operation determination unit is 0.5, and the dialog start state is S1. .

Based on the dialogue start state S1, the dialogue execution unit outputs the system response R1 "Please name a prefecture" to the user, and outputs the recognition target vocabulary V1 to the speech recognition unit to start the dialogue. .

When the user performs a voice input, the voice recognition unit performs a voice recognition process using the recognition target vocabulary V1, and outputs a recognition result and a score. For example, if the user inputs "Saga is", the recognition result will be "Shiga (0.88), Saga (0.87),
Kagawa (0.73), Kanagawa (0.52) and Kagoshima (0.50) "are output. The numerical value in parentheses is a score for each recognition candidate, and the closer to 1, the better the score.

When the recognition result is input, the transition destination dialog state determination operation determining unit refers to the transition table T1 defined in the current conversation state S1, and determines the hypothesis of the transition destination dialog state for the above five recognition results. As five dialogue states S15, S16,
Obtain S17, S18, S19. Next, because the score of the first-ranked recognition result “Shiga” is 0.88, which is above the threshold, it is decided to suspend the decision, and the five transition destination dialog state hypotheses are output to the provisional transition destination dialog state determination unit.

The provisional transition destination dialog state determination unit selects one provisional transition destination conversation state hypothesis from the input transition destination conversation state hypothesis based on the score of the recognition result. For example, the above five transition destination dialog state hypotheses S15, S16, S17, S18, S1
With respect to 9, the transition destination dialog state hypothesis S16 for “Shiga” having the best score is selected and output to the dialog operation execution unit. In addition, all transition destination dialog state hypotheses are added to the dialog state transition storage unit. For example, FIG. 3 shows a result of adding five transition destination dialog state hypotheses S15, S16, S17, S18, and S19 to the dialog state transition storage unit.

From the temporary transition destination dialog state determination unit, the dialog state S1
When 6 is input, the interactive operation execution unit returns the system response R16.
Along with outputting "Please enter the city name" to the user, the recognition target vocabulary V16 defined in the dialog state S16 and the four transition destination dialog state hypotheses S15, S17, and S1 stored in the dialog state transition storage unit.
8. The recognition target words V15, V17, V18, and V19 defined in S19 are output to the speech recognition unit.

[0028] In response to the system response "Please enter the city name" output by the interactive operation execution unit, the user is "Imari city."
, The voice recognition unit recognizes the vocabulary V15, V16, V1
Speech recognition processing was performed using 7, V18, and V19, and the recognition results were "Imari (0.91), Izumi (0.76), Isehara (0.30), Yokaichi
(0.11), Hiratsuka (0.09) ".

The destination dialog state determination operation determining unit refers to the transition tables T15, T16, T17, T18, and T19 defined in the destination dialog state hypotheses S15, S16, S17, S18, and S19, and determines the recognition result. Five dialog states S152, S153, S163, S182, and S192 are obtained as hypotheses of the transition destination dialog state. Since the score of "Yokaichi", which is the recognition result for S163, which is the transition destination dialog state hypothesis from the current dialog state S16, is 0.11 and less than the threshold,
The destination dialog state determination operation determination unit determines to determine the destination dialog state, and determines the destination dialog state hypotheses S152, S153, and S16.
3. Output S182 and S192 to the transition destination dialog state determination unit.

When the hypothesis of the transition destination dialog state is input, the transition destination dialog state determination unit confirms the user in order from, for example, a recognition result having a good score, and determines the transition destination dialog state.
If the destination dialogue state hypothesis S152, S153, S163, S182, S192 is entered, the user first asks, "Is it Imari City?"
When the user responds “Yes” to this, the transition destination dialog state is determined in S182.

After the transition destination dialog state is determined, the transition destination dialog state determination unit outputs the determined dialog state S182 to the dialog operation execution unit, writes the dialog state S182 into the dialog state transition storage unit, and furthermore, the dialog state transition The transition destination dialog state hypotheses S15, S16, S17, and S19 stored in the storage unit are deleted from the dialog state transition storage unit. The dialog state transition storage unit after performing the above operation is as shown in FIG.

When the dialogue state S182 is input, the dialogue execution unit outputs to the user the system response R182 "Please enter the town name" and also outputs the recognition target vocabulary V182 defined in the dialogue state S182 to the speech recognition unit. And continue the dialogue.

With the above operations, the dialog state transition storage unit holds a plurality of dialog state transition hypotheses in response to a user's input, and the transition destination dialog state determination operation determination unit determines that the recognition score for the currently selected hypothesis is a threshold. When the state becomes worse, the transition destination dialog state is determined as one, so that the user can be confirmed for each utterance and the recognition rate can be improved without having to proceed with the dialog steadily. As a result, a natural dialog between the user and the device can be realized, and the convenience of the user is improved.

Note that, from the configuration of the first embodiment, the dialog state transition storage unit, the transition destination dialog state determination unit, and the provisional transition destination dialog state determination unit are omitted, and the speech recognition unit, the dialog procedure storage unit, It is also possible to comprise a transition destination dialog state determination operation determination unit and a dialog operation execution unit. In this spoken dialogue apparatus, the dialogue procedure storage unit defines and stores a vocabulary to be recognized in each dialogue state, a system response, an answer assumed for the system response, and a transition destination dialogue state corresponding to the answer.

The speech recognition unit performs speech recognition on the input speech using the recognition target vocabulary corresponding to each conversation state stored in the conversation procedure storage unit, and outputs a plurality of recognition results. The destination dialog state determination operation determining unit determines the destination dialog state based on the recognition result from the speech recognition unit and the contents of the dialog procedure storage unit, and when the hypothesis of the destination dialog state satisfies a predetermined condition. Is determined to be one, and if the predetermined condition is not satisfied, the determination is suspended and the transition destination dialog state hypothesis is output. The dialogue execution unit outputs a system response for confirming the recognition result of the transition destination dialog state hypothesis from the transition destination dialog state determination operation determination unit when the hypothesis is determined to be one. Outputs the system response of the previous conversation state hypothesis.

In the above-structured speech dialogue apparatus, when the user performs a speech input, the speech recognition unit performs speech recognition processing using the recognition target vocabulary V1 stored in the dialogue procedure storage unit. Output recognition result. For example, if the user inputs "Saga", the recognition result is "Shiga (0.88), Saga (0.87), Kagawa (0.73), Kanagawa (0.52), Kagoshima (0.50)"
Output the five candidates. The numerical value in parentheses is the score for each recognition candidate as a predetermined condition and is 1
The closer to, the better the score.

When the recognition result is input, the transition destination dialog state determination operation determination unit refers to the transition table T1 defined in the current conversation state S1, and determines the hypothesis of the transition destination dialog state for the above five recognition results. As five dialogue states S15, S16,
Obtain S17, S18, S19 and set flags. Next, assuming that a predetermined condition is a score of a recognition result,
Since the score of the first-ranked recognition result "Shiga" is 0.88, which is above the threshold value of the predetermined condition, it is decided to suspend the determination.

Next, the transition destination dialog state determination operation determining unit selects one temporary transition destination dialog state hypothesis from the transition destination dialog state hypothesis based on the score of the recognition result. For example, the above five transition destination dialog state hypotheses S15, S16, S17, S18, S19
With respect to, the transition destination dialog state hypothesis S16 for “Shiga” having the best score is selected and output to the dialog operation execution unit.

The dialogue operation execution part is a transition destination dialogue state hypothesis S16.
The system response R16 "Please enter the city name" is output to the user.

In response to the system response "Please enter the city name" output by the interactive operation execution unit, the user is "Imari city."
When the speech recognition unit is input, the speech recognition unit performs speech recognition processing using the recognition target vocabulary V15, V16, V17, V18, and V19 of the dialogue procedure storage unit, and as a recognition result, "Imari (0.91), Izumi (0.76), Isehara (0.30), Yokaichi (0.11), Hiratsuka (0.09) ".

The destination dialog state determination operation determination unit refers to the transition tables T15, T16, T17, T18, and T19 defined in the destination dialog state hypotheses S15, S16, S17, S18, and S19, and determines the recognition result. Five dialog states S152, S153, S163, S182, and S192 are obtained as hypotheses of the transition destination dialog state.

Since the score of "Yokaichi", which is the recognition result for S163, which is the transition destination dialog state hypothesis from the current conversation state S16, is 0.11 or less than the threshold value of the predetermined condition, the transition destination dialog state determination operation determination unit determines the transition destination dialog state. Is determined.

The interactive operation execution unit confirms the user in order from the recognition result having the highest score, for example, and determines the transition destination interactive state. Transition destination dialog state hypothesis S152, S153, S163, S182,
When S192 is entered, the user is first asked, "Is it Imari City?" And the user responds "Yes"
The destination dialog state is determined in S182 by responding with.

When the dialogue state S182 is input, the dialogue execution unit outputs the system response R182 "Please enter the town name" to the user, and also outputs the recognition target vocabulary V182 defined in the dialogue state S182 to the speech recognition unit. And continue the dialogue.
As described above, the operation is repeated until the transition destination dialog state determining operation determining unit determines one hypothesis of the transition destination dialog state.

Embodiment 2 The second embodiment is different from the above-described first embodiment in the operation of the transition destination dialog state deciding operation determining unit, and the other is the same as the above-described first embodiment. Hereinafter, the operations of the interaction procedure storage unit and the transition destination interaction state determination operation determination unit of FIG. 1 will be described.

FIGS. 2, 5 and 6 show examples of the dialogue state held in the dialogue procedure storage unit. Dialogue state S18 in FIG.
Vocabulary specified in 231, S18241, S18251, S18281 V1823
1, V18241, V18251, and V18281 are large in scale, and it is not desirable to perform speech recognition at the same time as the vocabulary to be recognized in other dialogue states, so it is necessary to determine the dialogue state before transitioning to these dialogue states There is a condition that there is.

Hereinafter, a case will be described in which, after the apparatus and the user start a conversation from the conversation state S1, the conversation is performed in the same manner as in the first embodiment, and the current conversation state is S182.

When the user inputs "Kurokawa is" in response to the system response "please name of town" output by the interactive action execution unit, the speech recognition unit performs speech recognition processing using the recognition target vocabulary V182. "Kurokawa (0.95), Okawa (0.88), Okawachi (0.70), Otsubo (0.11), Tachibana (0.03)" are output as recognition results.

The transition destination dialog state determination operation determining unit refers to the transition table T182 to determine the transition destination dialog state hypothesis S1825, S182.
2, S1823, S1824, S1828 are obtained. Next, since the score of the first recognition result “Kurokawa” is equal to or larger than the threshold, the determination of the transition destination dialog state determination operation is suspended, and the transition destination dialog state hypothesis is output to the provisional transition destination dialog state determination unit.

The provisional transition destination dialog state determination unit selects the transition destination dialog state hypothesis S1825 for “Kurokawa” having the best score and outputs it to the dialogue execution unit. Add to the storage unit. After these processes, the dialog state transition storage unit is as shown in FIG.

From the temporary transition destination dialog state determining unit, the dialog state S
When 1825 is entered, the interactive execution unit responds with a system response.
While outputting R1825 "Please type of business" to the user,
The recognition target vocabulary V1825 defined in the dialog state S1825,
Recognized vocabulary defined by the four transition destination dialog state hypotheses stored in the dialog state transition storage unit V1822, V1823, V1824, V
1828 is output to the voice recognition unit.

In response to the system response "Please enter the type of business,"
When the user inputs "Ryokan," the voice recognition unit performs voice recognition using the recognition target vocabulary V1825, V1822, V1823, V1824, and V1828, and as a recognition result, "Ryokan (0.9
5), barber (0.62), tourism (0.51), rheumatology (0.27),
Nursing home (0.10), hunting gun (0.02) ".

The destination dialog state determination operation determination unit refers to the transition tables T1825, T1822, T1823, T1824, and T1828, and
9 dialog states S18231, S18232, S18241, S18242, S18243, S18
251, S18252, S18281, S18282 are obtained. The score of the best recognition result “Ryokan” is 0.95, which is above the threshold, but the transition destination hypothesis for “Ryokan” is S18231, S18.
241, S18251, and S18281 are all dialogue states that require a definitive action, so the destination dialog state determination action deciding unit decides to determine the destination dialog state, and changes the destination dialog state hypothesis to the destination dialog state. Output to the state determination unit.

The destination dialog state determination unit confirms to the user, "Is the business type a ryokan?", And the user responds "Yes" to the destination dialog state to change the destination dialog state to S18231, S18.
241, S18251, and S18281, the transition destination dialog state hypothesis S1822 stored in the dialog state transition storage unit is deleted. Furthermore, the transition dialogue state hypothesis S1823, S1824, S1825, S1
Since 828 exists, the transition destination dialog state determination unit determines the hypothesis by confirming the town name to the user. First, the user confirms the recognition result for the dialogue state S1825 with the best score, "Is the town name Kurokawa?" The user responds “Yes” to this confirmation, and the dialogue state S182 is determined. As a result, the final transition destination conversation state is determined in S18251.

When the dialogue state S18251 is input, the dialogue execution unit outputs the vocabulary V18251 to be recognized to the speech recognition unit, and outputs the system response R18251 "What is the name of the inn" to the user to carry out the dialogue. continue.

According to the above operation, since the specified vocabulary to be recognized is large, it is not desirable to perform the speech recognition process simultaneously with the vocabulary to be recognized in another dialogue state. For the dialog state that needs to be changed, the transition destination dialog state deciding operation decision unit decides the execution of the definitive operation, and the transition destination dialog state deciding unit decides the transition destination dialog state. improves.

Embodiment 3 The third embodiment is different from the above-described first embodiment in the operation of the transition destination dialog state determining operation determining unit, and is otherwise the same as the above-described first embodiment. Hereinafter, regarding the operation of the transition destination dialog state determination operation determining unit in FIG. 1, the dialog state stored in the dialog procedure storage unit is FIG. 8, the telephone number database is FIG. 9, and the number N of recognition results output by the speech recognition unit is The case of 3 will be described as an example.

Based on the dialogue start state S1, the dialogue execution unit responds with the system response R1 "Which phone number are you looking for?"
Is output to the user and the vocabulary V1 to be recognized is output to the speech recognition unit to start the dialogue.

When the user performs a voice input, the voice recognition unit performs a voice recognition process using the recognition target vocabulary V1, and outputs a recognition result and a score. For example, if the user inputs "Amataro", the recognition result is "Amata (0.88),
(0.87), Amagi (0.73) "are output.

When the recognition result is input, the transition destination dialog state determination operation determining unit obtains S2, S3, S4 as the transition destination dialog state hypothesis with reference to the transition table T1. Next, it is checked whether or not the item value for the uninput item is uniquely determined by determining the recognition result from the voice recognition unit. In the present embodiment, the input items of the user are a prefecture name, a municipal name, a business type, and a target name. At the moment, only the target name has been input, and it is sufficient to check whether or not an uninput item is uniquely determined from only the target name obtained as a recognition result by referring to the telephone number database of FIG. . For the recognition candidate “Amata”, data number 5,
Although there are two types of data, the prefecture name, which has not been entered, cannot be uniquely determined just by confirming the recognition result. The same applies to the other recognition results "Amataro" and "Amagi".
Since uninput items are not uniquely determined for all recognition candidates, it is determined that the determination of the transition destination dialog state is to be suspended,
The three transition destination dialog state hypotheses are output to the provisional transition destination dialog state determination unit.

The provisional transition destination dialog state determining unit determines, for example, the transition destination dialog state hypothesis S3 for “Amata” having the highest score.
Is selected and output to the interactive operation execution unit. Furthermore, all the transition destination dialog state hypotheses are added to the dialog state transition storage unit.

From the temporary transition destination dialog state determination unit, the dialog state S3
Is input, the dialogue execution unit outputs the system response R3 "What prefecture is it?"
The recognition target vocabulary V3 specified in S3 and the recognition target vocabulary V2 and V4 specified in the two transition destination dialog state hypotheses S2 and S4 stored in the dialog state transition storage unit are output to the speech recognition unit.

When the user inputs "Kanagawa prefecture" in response to the system response "What prefecture is it?" Output by the dialogue execution unit, the speech recognition unit causes the recognition target vocabulary V2, V
3. Perform speech recognition processing using V4, and output "Kanagawa (0.95), Kagawa (0.72), Saga (0.41)" as the recognition result.

The transition destination dialog state determination operation determination unit refers to the transition tables T2, T3, and T4, and as a hypothesis of the transition destination dialog state with respect to the recognition result, the nine dialog states S22, S23, S24,
Obtain S32, S33, S34, S42, S43, S44. Next, it is checked whether or not the item value for the uninput item is uniquely determined by determining the recognition result from the voice recognition unit. At this point, the subject name and prefecture name have been entered. Referring to the telephone number database in FIG. 9, the data with the name “Amata” and the prefecture name “Kanagawa” are two data numbers 5 and 6. However, the city name, which is an uninput item, is not uniquely determined just by determining the prefecture name. Similarly, for the data 1, 2, and 3 with the name “Amataro” and the prefecture name “Kanagawa”, it is determined that the determination of the state of the transition destination dialogue is suspended because the city names are not uniquely determined, and the nine transition destination dialogues are determined. The state hypothesis is output to the provisional transition destination dialog state determination unit.

The provisional transition destination dialog state determination unit selects the transition destination dialog state hypothesis S32 for "Kanagawa" having the highest score from the current conversation state S2, and outputs it to the dialog operation execution unit. In addition, all nine transition destination dialog state hypotheses are added to the dialog state transition storage unit.

From the temporary transition destination dialog state determination unit, the dialog state S3
When 2 is entered, the interactive execution unit returns the system response R32
"What city is it?"
The recognition target vocabulary V32 defined in S32 and the eight transition destination dialog state hypotheses stored in the dialog state transition storage unit S22, S42, S2
3. Output the recognition target vocabulary V22, V42, V23, V33, V43, V24, V34, V44 specified in S33, S43, S24, S34, S44 to the voice recognition unit.

When the user inputs "Kamakura City" in response to the system response "What city is it?" Output by the dialogue execution part, the speech recognition part makes the recognition target vocabulary V22, V32, V4.
2, V23, V33, V43, V24, V34, V44 are used for speech recognition processing, and the results of the recognition are “Kamakura (0.87), Kawasaki (0.66),
Karatsu (0.28) "is output.

The transition destination dialog state determination operation determining unit determines the transition table T22, T32, T42, T23, T33, T43, T24, T34, T44.
, Nine dialog states S222, S223, S322, S323, S422, S4
Obtain 23, S243, S343, S443. Next, it is checked whether or not the item value for the uninput item is uniquely determined by determining the recognition result from the voice recognition unit. At this point, the target name, the prefecture name, and the city name have been input. Referring to the telephone number database of FIG. 9, there is no data of the name “Amata”, the prefecture name “Kanagawa”, and the city name “Kamakura”.

On the other hand, the name “Amataro”, the prefecture name “Kanagawa”,
For the city name "Kamakura", there is data with data number 1, and by determining the city name, the unentered items such as the town name and business type are uniquely determined. Therefore, it is determined that the transition destination dialog state is determined, and the transition destination dialog state hypothesis S222, S
223, S322, S323, S422, S423, S243, S343, S443 are output to the transition destination dialog state determination unit.

When the hypothesis of the transition destination dialog state is input, the transition destination dialog state determination unit determines the transition destination dialog state in S223 by confirming “Kamakura” which is the recognition result having the highest score, and The conversation state S223 is output to the operation execution unit.

When the dialogue state execution unit S223 is input, the dialogue operation execution unit stores the dialogue transition sequence stored in the dialogue state transition storage unit and the item values for all input items from the telephone number database shown in FIG. ”, Prefecture name“ Kanagawa ”, city name“ Kamakura ”, town name“ Ofuna ”, and business type“ Izakaya ”. Since the telephone number“ 0467-00-0000 ”is uniquely determined, the telephone number is returned to the user. I do.

With the above operation, a plurality of hypotheses of the dialog state transition in response to the user's input are held, and even if all the input items from the user have not been input, the recognition result is determined and the items corresponding to the non-input items are determined. When the value is uniquely determined, the transition destination dialog state is determined as one, so the confirmation rate can be improved without confirming the user for each utterance and proceeding definitively, and the number of confirmation dialogs Therefore, a natural conversation between the user and the device can be realized, and the convenience of the user is improved.

Embodiment 4 Embodiment 4 is different from Embodiment 1 described above in that the operation of the transition destination dialog state deciding operation determining unit is different, and the rest is the same as Embodiment 1 described above.
Hereinafter, the operation of the transition destination dialog state determination operation determining unit in FIG. 1 will be described by taking as an example a case where the dialog state in FIG. 10 is held in the dialog procedure storage unit.

Based on the dialogue start state S1, the dialogue execution unit outputs the system response R1 "Please name the prefecture" to the user, and outputs the recognition target vocabulary V1 to the speech recognition unit to start the dialogue. .

When the user makes a speech input, the speech recognition unit performs a speech recognition process using the recognition target vocabulary V1, and outputs a recognition result and a score. For example, if the user inputs “Saga is”, the recognition result will be “Saga (0.92), Shiga (0.80),
Kagawa (0.73), Kanagawa (0.52) and Kagoshima (0.50) "are output.

When the recognition result is input, the transition destination dialog state determining operation determining unit refers to the transition table T1 defined in the current conversation state S1, and determines the hypothesis of the transition destination dialog state for the above five recognition results. 5 conversation states S15, S16, S1
Get 7, S18, S19. Next, the first recognition result “Saga” has a score of 0.92, which is equal to or higher than the threshold, so that it is decided to suspend the decision, and all five transition destination dialog state hypotheses having a common system response are provisionally transitioned to the transition destination dialog state determination unit. Output to

The provisional transition destination dialog state determination unit selects the transition destination dialog state hypothesis S18 for “Saga” having the best score, outputs it to the dialogue execution unit, and stores all transition destination dialog state hypotheses in the dialog state transition storage unit. Add to

From the temporary transition destination dialog state determination unit, the dialog state S
When "18" is input, the interactive action execution unit returns the system response R1.
8 Output “Please enter the city name” to the user, the vocabulary V18 to be recognized, and the four target dialogue state hypotheses S15, S16, S17, S19 stored in the dialogue state transition storage unit. Vocabulary V15, V16, V17, V19 are output to the speech recognition unit.

In response to the system response "Please enter the city name" output by the interactive operation execution unit, the user is "Imari city."
, The speech recognition unit recognizes the vocabulary V15, V16,
Speech recognition processing is performed using V17, V18, and V19, and "Imari (0.91), Izumi (0.76), Isehara (0.30), Yokaichi (0.11), Hiratsuka (0.09)" are output as recognition results.

The transition destination dialog state determination operation determination unit refers to the transition tables T15, T16, T17, T18, and T19, and as a hypothesis of the transition destination dialog state for the recognition result, the five dialog states S1
52, S153, S163, S182, S192 are obtained. Although the score of the best recognition result "Imari" is equal to or higher than the threshold, the destination dialog state hypothesis S152, S153, S163, S182, and S192 do not have a common system response. Determines that the destination dialog state is determined, and outputs the destination dialog state hypotheses S152, S153, S163, S182, and S192 to the destination dialog state determination unit.

The transition destination dialog state determination unit operates in the same manner as in the first embodiment, determines the transition destination dialog state in S182, and outputs it to the dialog operation execution unit. When the transition destination dialog state S182 is input, the dialog operation execution unit operates in the same manner as in the first embodiment, and continues the dialog with the user.

With the above operation, the dialog state transition storage section holds a plurality of hypotheses of the dialog state transition in response to the user's input, and the transition destination dialog state determining operation determining section determines that the system utterance common to the transition destination dialog state hypothesis is generated. When the dialogue no longer exists, the transition destination dialogue state is determined as one, so the user can be confirmed for each utterance and the recognition rate can be improved without having to proceed with the dialogue steadily. As a result, a natural dialog between the user and the device can be realized, and the convenience of the user is improved.

Fifth Embodiment The fifth embodiment is different from the first embodiment in the operation of the transition destination dialog state determining operation determining unit, and the other is the same as the first embodiment. Hereinafter, the operation of the transition destination dialog state deciding operation determining unit in FIG. 1 will be described by taking as an example a case where the dialog state in FIG. 11 is held in the dialog procedure storage unit. In the example shown in FIG. 11, in the conversation state S152, a plurality of system responses R152-1 "What is the town of Isehara city" and R152-2 "Please name the town" are specified.

First, in the same manner as in the fourth embodiment, the dialogue is started from the dialogue start state S1, the dialogue state is reached, and the user responds to the system response R18 “Please enter the city name” with “Imari city”. Then, the voice recognition unit returns the recognition result `` Imari (0.9
1), Izumi (0.76), Isehara (0.30), Yokaichi (0.11), Hiratsuka
(0.09) "will be described.

The transition destination dialog state determination operation determining unit refers to the transition tables T15, T16, T17, T18, and T19 and, as in the fourth embodiment, determines five transition states as the hypothesis of the transition destination dialog state for the recognition result. Obtain S152, S153, S163, S182, S192. The score of the best recognition result "Imari" is equal to or greater than the threshold value, and the destination dialog state hypothesis S152, S153,
Since there is a system response common to S163, S182, and S192, "Please enter the town name," the transition destination dialog state determination operation determination unit determines to suspend the determination of the transition destination dialog state, and the transition destination dialog state hypothesis S152, S153, S163, S182, and S192 are output to the provisional transition destination dialog state determination unit.

When the transition destination dialog state hypothesis S152, S153, S163, S182, S192 is input, the provisional transition destination dialog state determination unit changes the dialog state S182 for the recognition result “Imari” having the highest score to the provisional transition destination dialog state. And outputs it to the interactive operation execution unit.

When the dialogue state execution unit S182 inputs the dialogue state S182, the dialogue execution unit hypotheses the destination dialogue state hypotheses S152, S153, S163, and S163 from among a plurality of system responses defined in the dialogue state S182.
Output R182-2 "Please enter the town name" common to the system response specified in S192 as the system response and continue the dialogue.

On the other hand, as in the first embodiment, the dialogue is started from the dialogue start state S1, the dialogue state is reached, and the user responds to the system response R16 “Please enter the city name” with “Imari city”. Therefore, the case where the transition destination dialog state determination unit confirms with the user and determines the transition destination dialog state to S182 will be described.

When the dialogue state S182 is input, the dialogue operation execution unit executes the system response R18 specified in the dialogue state S182.
2-1 Of "What town in Imari city" and R182-2 "Please name the town", for example, the first defined R182-1
Is output as a system response and the dialogue is continued.

By describing a plurality of system responses in each dialogue state stored in the dialogue procedure storage unit by the above operation, when a common system utterance is present in the transitional dialogue state hypothesis, the transitional dialogue state is described. When the finalized action deciding unit does not perform the finalized action by confirmation, outputs a common system utterance to each transition destination dialog state hypothesis and continues the dialogue, while the transition destination dialog state determinator determines the transition destination dialog state Since the system response unique to the confirmed dialog state can be performed, the recognition rate can be improved without confirming the user every utterance and proceeding deterministically. Since a natural response can be made, a natural conversation between the user and the device can be realized, and the convenience of the user is improved.

Embodiment 6 FIG. The sixth embodiment is different from the above-described first embodiment in the operation of the transition destination dialog state deciding operation determining unit, and is otherwise the same as the above-described first embodiment. Hereinafter, the operation of the transition destination dialog state determining operation determining unit in FIG. 1 will be described by taking the case where the dialog state of FIGS. A case in which the vocabulary scale threshold used for determination is 300 will be described.

Based on the dialogue start state S1, the dialogue execution unit outputs the system response R1 "Please name the prefecture" to the user, and outputs the recognition target vocabulary V1 to the speech recognition unit to start the dialogue. .

When the user performs voice input, the voice recognition unit performs voice recognition processing using the recognition target vocabulary V1, and outputs a recognition result and a score. For example, if the user inputs “Saga is”, the recognition result will be “Saga (0.92), Shiga (0.80),
Kagawa (0.73), Kanagawa (0.52) and Kagoshima (0.50) "are output.

When the recognition result is input, the transition destination dialog state determining operation determining unit refers to the transition table T1 defined in the current conversation state S1, and determines the hypothesis of the transition destination dialog state for the above-described five recognition results. 5 conversation states S15, S16, S1
Get 7, S18, S19. Next, the vocabulary obtained by summing all the recognition target vocabularies V15, V16, V17, V18, and V19 of the transition destination dialog state hypothesis is obtained. V15, V16, V17, V18, V19 are respectively
Because the names of Kanagawa, Shiga, Kagawa, Saga, and Kagoshima prefectures, the total vocabulary is 5
The vocabulary size is 52 because the names of all the cities in the prefectures are not the same in these five prefectures.

Since this is smaller than the threshold value of 300, the transition destination dialog state determination operation determination unit determines to suspend the determination of the transition destination dialog state hypothesis, and tentatively determines all five transition destination dialog state hypotheses having a common system response. Output to the transition destination dialog state determination unit.

The provisional transition destination dialog state determination unit selects the transition destination dialog state hypothesis S18 for “Saga” having the best score, outputs it to the dialogue execution unit, and stores all transition destination dialog state hypotheses in the dialog state transition storage unit. Add to

From the provisional transition destination dialog state determination unit, the dialog state S
When "18" is input, the interactive action execution unit returns the system response R1.
8 While outputting “Please enter the city name” to the user, the recognition target vocabulary V18 and the four transition destination dialog state hypotheses S15, S16, S17, and S19 stored in the dialog state transition storage unit are recognized. Vocabulary V15, V16, V17, V19 are output to the speech recognition unit.

In response to the system response "Please enter the city name" output by the interactive operation execution unit, the user is "Imari city."
, The speech recognition unit recognizes the vocabulary V15, V16,
Speech recognition processing is performed using V17, V18, and V19, and "Imari (0.91), Izumi (0.76), Isehara (0.30), Yokaichi (0.11), Hiratsuka (0.09)" are output as recognition results.

The destination dialog state determination operation determining unit refers to the transition tables T15, T16, T17, T18, and T19, and as a hypothesis of the destination dialog state for the recognition result, the five dialog states S1
52, S153, S163, S182, S192 are obtained. Next, all the target vocabularies V152, V153, V163, V
Find the vocabulary of the sum of 182 and V192. V152, V153, V163,
V182 and V192 are the names of the towns of Isehara, Kanagawa, Hiratsuka, Kanagawa, Yokaichi, Shiga, Imari, Saga, and Izumi, Kagoshima, respectively. Are all the street names. The total number is 332
However, since the same town name exists in these five different cities, the number of words is 327 different.

Since this is larger than the threshold value of 300, the transition destination dialog state determination operation determination unit determines to determine the transition destination dialog state, and the transition destination dialog state hypothesis S152, S153, S163, S
182 and S192 are output to the transition destination dialog state determination unit.

The transition destination dialog state determination unit operates in the same manner as in the first embodiment, determines the transition destination dialog state in S182, and outputs it to the dialog operation execution unit. When the transition destination dialog state S182 is input, the dialog operation execution unit operates in the same manner as in the first embodiment, and continues the dialog with the user.

By the above operation, when the size of the vocabulary to be recognized specified in the transition destination dialog state hypothesis is large and the recognition rate may be reduced, the transition destination dialog state determination operation determination unit determines the execution of the determination operation. Since the destination dialog state determination unit determines the destination dialog state, the vocabulary to be recognized can be limited and the recognition rate can be improved.

Embodiment 7 FIG. The seventh embodiment is different from the first embodiment in the operation of the transition destination dialog state determination operation determining unit, and the other is the same as the first embodiment. Hereinafter, regarding the operation of the transition destination dialog state determination operation determining unit of FIG. 1, the dialog state of FIG. 2 and FIG. 5 is held in the dialog procedure storage unit, and the transition destination dialog state determination operation determination unit uses for the determination operation determination. The case where the threshold of the length of the transition sequence is 2 will be described.

On the basis of the dialogue start state S1, the dialogue execution unit outputs the system response R1 "Please name the prefecture" to the user, and outputs the recognition target vocabulary V1 to the speech recognition unit to start the dialogue. .

When the user makes a voice input, the voice recognition unit performs a voice recognition process using the recognition target vocabulary V1, and outputs a recognition result and a score. For example, if the user inputs “Saga is”, the recognition result will be “Saga (0.92), Shiga (0.80),
Kagawa (0.73), Kanagawa (0.52) and Kagoshima (0.50) "are output.

When the recognition result is input, the transition destination dialog state determination operation determination unit refers to the transition table T1 defined in the current conversation state S1, and determines the hypothesis of the transition destination dialog state for the above five recognition results. 5 conversation states S15, S16, S1
Get 7, S18, S19. The transition destination dialog state determination operation determination unit obtains the length of the transition sequence of the transition destination dialog state hypothesis by referring to the dialog state transition storage unit. At this point, the conversation start state S
Since no dialog state hypothesis after 1 is stored in the dialog state transition storage unit, the length of the transition sequence is 0, which is smaller than the threshold value of 2. Therefore, the transition destination dialog state determination operation determination unit determines to suspend the determination of the transition destination dialog state, and outputs all five transition destination dialog state hypotheses to the provisional transition destination dialog state determination unit.

The provisional transition destination dialog state deciding section selects the transition destination dialog state hypothesis S18 for “Saga” having the best score, outputs it to the dialogue execution section, and stores all transition destination dialog state hypotheses in the dialog state transition storage section. Add to As a result of the above operation, the contents of the dialog state transition storage unit are as shown in FIG.

From the temporary transition destination dialog state determining unit, the dialog state S
When "18" is input, the interactive action execution unit returns the system response R1.
8 Output “Please enter the city name” to the user, recognize the vocabulary V18 to be recognized, and the four transition destination hypotheses S15, S16, S17, S19 stored in the dialog state transition storage unit. Vocabulary V15, V16, V17, V19 are output to the speech recognition unit.

In response to the system response "Please enter the city name" output by the interactive operation execution unit, the user is "Imari city."
, The speech recognition unit recognizes the vocabulary V15, V16,
Speech recognition processing is performed using V17, V18, and V19, and "Imari (0.91), Izumi (0.76), Isehara (0.30), Yokaichi (0.11), Hiratsuka (0.09)" are output as recognition results.

The destination dialog state determination operation determining unit refers to the transition tables T15, T16, T17, T18, and T19, and as a hypothesis of the destination dialog state for the recognition result, the five dialog states S1
52, S153, S163, S182, S192 are obtained. Next, referring to the contents of the dialog state transition storage unit shown in FIG. 12, the length of the transition sequence from the dialog start state S1 to the current dialog state S18 is 1, which is smaller than the threshold value of 2. Therefore,
The transition destination dialog state determination operation determination unit determines that the determination of the transition destination dialog state is suspended, and outputs all five transition destination dialog state hypotheses to the provisional transition destination dialog state determination unit.

When the transition destination dialog state hypothesis S152, S153, S163, S182, and S192 is input, the provisional transition destination dialog state determination unit changes the dialog state S182 for the recognition result “Imari” having the highest score to the provisional transition destination dialog state. And outputs it to the interactive operation execution unit. Further, all the transition destination dialog state hypotheses are added to the dialog state transition storage unit, and the contents of the dialog state transition storage unit are as shown in FIG.

From the temporary transition destination dialog state determination unit, the dialog state S
When 182 is entered, the interactive execution unit returns the system response R
182 Outputs "Please enter the town name" to the user, and stores the vocabulary V182 to be recognized and the dialog state transition storage unit.
The recognition target vocabulary V152, V153, V163, and V192 defined in the four transition destination dialog state hypotheses S152, S153, S163, and S192 are output to the speech recognition unit.

When the user inputs "Kurokawa is" in response to the system response "Please enter the town name" output by the interactive operation execution unit, the speech recognition unit uses the recognition target vocabulary V182, V152, V
Speech recognition was performed using 153, V163, and V192, and the results of recognition were "Kurokawa (0.90), Hirokawa (0.64), Okawa (0.42), Kushibashi
(0.13), Kurobe-oka (0.11) ".

The transition destination dialog state determination operation determining unit refers to the transition table T182 to determine the transition destination dialog state hypothesis S1825, S182.
2, S1823, S1824, S1828 are obtained. Next, referring to the contents of the dialog state transition storage unit shown in FIG. 13, the length of the transition sequence from the dialog start dialog state S1 to the current dialog state S182 is 2, which is equal to the threshold value. Therefore, the transition destination dialog state determination operation determination unit determines to determine the transition destination dialog state, and outputs five transition destination dialog state hypotheses to the transition destination dialog state determination unit.

When the transition destination dialog state hypothesis S1825, S1822, S1823, S1824, S1828 is input, the transition destination dialog state determination unit confirms the user with the highest score “Kurokawa” to determine the transition destination dialog state. Determined in S1825, and output to interactive operation execution unit.

When the dialogue state S1825 is input, the dialogue operation execution unit outputs the recognition target vocabulary V1825 to the speech recognition unit, and outputs the system response R1825 “Please type of business” to the user to continue the dialogue.

With the above operation, the dialog state transition storage unit holds a plurality of hypotheses of the dialog state transition in response to the user's input, and the transition destination dialog state determining operation determining unit determines the dialog state from the most recently determined dialog state. When the length of the transition hypothesis sequence exceeds the threshold, the transition destination dialog state is determined as one, so the user is confirmed for each utterance and the recognition rate is improved without having to proceed deterministically. Since the number of confirmation dialogues can be reduced, a natural dialogue between the user and the device can be realized, and the convenience of the user improves.

[0118]

【The invention's effect】

As described above, according to the present invention, when the currently selected hypothesis satisfies a predetermined condition, the transition destination dialog state determination operation determination unit determines one transition destination dialog state. The recognition rate can be improved without confirming to the user for each utterance and proceeding deterministically, and the number of confirmation dialogues can be reduced to realize a natural dialogue between the user and the device, which is convenient for the user. The performance is improved.

Further, according to the present invention, the dialog state transition storage unit holds a plurality of hypotheses of the dialog state transition in response to the user's input, and the transition destination dialog state determination operation determination unit recognizes the currently selected hypothesis. When the score becomes worse than the threshold, the transition destination dialog state is determined as one, so the user can be confirmed for each utterance, and the recognition rate can be improved without having to proceed deterministically. Since the number of times is reduced, a natural conversation between the user and the device can be realized, and the convenience of the user is improved.

Further, according to the present invention, since the specified vocabulary to be recognized is large, it is not desirable to perform the speech recognition process simultaneously with the vocabulary to be recognized in another dialogue state. For the dialog state that needs to be performed, the destination dialog state deciding operation deciding unit decides the execution of the definitive operation, and the transition destination dialog state deciding unit decides the transition destination dialog state. The recognition rate is improved.

Further, according to the present invention, a plurality of hypotheses of a dialog state transition in response to a user's input are held, and even if all the input items from the user have not been input, the recognition result is determined so that no input is made. When the item value for an item is uniquely determined, the transition destination dialog state is determined as one, so the user can be confirmed for each utterance and the recognition rate can be improved without steadily proceeding with the dialogue. Since the number of conversations is reduced, a natural conversation between the user and the device can be realized, and the convenience of the user is improved.

Further, according to the present invention, the conversation state transition storage unit holds a plurality of conversation state transition hypotheses in response to a user's input, and the transition destination conversation state determination operation determining unit determines the common state of the transition destination conversation state hypothesis. When the system utterance no longer exists, the transition destination dialog state is determined as one, so the user can be confirmed for each utterance and the recognition rate can be improved without steadily proceeding with the dialog. Since the number of times is reduced, a natural conversation between the user and the device can be realized, and the convenience of the user is improved.

Further, according to the present invention, by describing a plurality of system responses in each dialog state stored in the dialog procedure storage unit, when a common system utterance exists in the transition destination dialog state hypothesis, the transition state is determined. The destination dialog state determination operation deciding unit does not perform the confirmation operation by confirmation, outputs a system utterance common to each transition destination dialog state hypothesis, and continues the dialogue, while the transition destination dialog state determination unit determines the transition destination dialog state. If confirmed,
Since a system response specific to the confirmed dialogue state can be performed,
The recognition rate can be improved without confirming the user every utterance and proceeding deterministically, and a natural response can be made according to the dialog state transition. It can be realized and user convenience is improved.

Further, according to the present invention, when the size of the recognition target vocabulary specified in the transition destination dialog state hypothesis is large and the recognition rate may be reduced, the transition destination dialog state determination operation determining unit executes the determination operation. Is determined, and the transition destination dialog state determination unit determines the transition destination dialog state, so that the recognition target vocabulary can be limited and the recognition rate can be improved.

Further, according to the present invention, the dialog state transition storage unit holds a plurality of hypotheses of the dialog state transition in response to the user's input, and the transition destination dialog state determining operation determining unit determines the most recently determined dialog state from the most recently determined dialog state. If the length of the dialogue state transition hypothesis sequence exceeds the threshold, the transition destination dialogue state is determined as one. Rate can be improved, and the number of confirmation dialogues can be reduced, so that a natural dialogue between the user and the device can be realized, thereby improving the convenience of the user.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a voice interaction device according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an example of a conversation state held in a conversation procedure storage unit according to the first embodiment;

FIG. 3 is an explanatory diagram of a result of adding a transition destination dialog state hypothesis according to the first embodiment to a dialog state transition storage unit.

FIG. 4 is an explanatory diagram of a dialog state transition storage unit of a transition destination dialog state hypothesis determination result in the first embodiment.

FIG. 5 is an explanatory diagram illustrating an example of a dialogue state stored in a dialogue procedure storage unit according to the second embodiment.

FIG. 6 is an explanatory diagram showing an example of a conversation state held in a conversation procedure storage unit according to the second embodiment.

FIG. 7 is an explanatory diagram of a result of adding a transition destination dialog state hypothesis according to the second embodiment to a dialog state transition storage unit.

FIG. 8 is an explanatory diagram of a conversation state stored in a conversation procedure storage unit according to the third embodiment.

FIG. 9 is an explanatory diagram of a telephone number database according to the third embodiment.

FIG. 10 is an explanatory diagram of a conversation state stored in a conversation procedure storage unit according to the fourth embodiment.

FIG. 11 is an explanatory diagram of a dialogue state stored in a dialogue procedure storage unit according to the fifth embodiment.

FIG. 12 is an explanatory diagram of a result of adding a transition destination dialog state hypothesis according to the seventh embodiment to a dialog state transition storage unit.

FIG. 13 is an explanatory diagram of a transition sequence from a conversation start conversation state to a current conversation state in the conversation state transition storage unit according to the seventh embodiment.

FIG. 14 is a configuration diagram of a conventional recognition candidate extraction device.

[Explanation of symbols]

1: Speech recognition unit, 2: Dialogue procedure storage unit, 3: Dialogue state transition storage unit, 4: Transition destination dialogue state determination operation determination unit, 5: Transition destination dialogue state determination unit, 6: Temporary transition destination dialogue state determination unit , 7: Interactive operation execution unit.

 ────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yasushi Ishikawa 2-3-2 Marunouchi, Chiyoda-ku, Tokyo F-term (reference) in Mitsubishi Electric Corporation 5D015 AA05 JJ00 LL11

Claims (10)

[Claims] A speech recognition unit, a conversation procedure storage unit, a transition destination conversation state determination operation decision unit, and a conversation operation execution unit, and a speech for acquiring information required by a user through speech conversation. A dialogue apparatus, wherein a dialogue procedure storage unit defines and stores a vocabulary to be recognized in each dialogue state, a system response, an answer assumed for the system response, and a transition destination dialogue state corresponding to the answer, and performs speech recognition. The unit performs speech recognition on the input speech using the recognition target vocabulary corresponding to each dialogue state stored in the dialogue procedure storage unit, outputs a plurality of recognition results, and determines a transition destination dialogue state determination operation. The unit determines a transition destination conversation state based on the recognition result from the speech recognition unit and the contents of the conversation procedure storage unit, and determines one when the hypothesis of the transition destination conversation state satisfies a predetermined condition, If the conditions are not met When the decision is made to suspend the decision and the transition destination dialog state hypothesis is output, the dialogue execution unit determines the transition destination dialog state hypothesis from the transition destination dialog state decision operation decision unit when the hypothesis is decided to be one. And outputting a system response of a transition destination dialog state hypothesis when outputting a system response for confirming the confirmation. 2. A dialog state transition storage unit, a transition destination dialog state determination unit, and a provisional transition destination dialog state determination unit are added, and the transition destination dialog state determination operation determination unit determines the recognition result from the speech recognition unit. Determines whether the hypothesis of the destination dialog state determined from the contents of the dialog state transition storage unit or the dialog procedure storage unit is determined or suspended, outputs the destination dialog state hypothesis, and outputs the destination dialog state. The state determination unit inputs the transition destination dialog state hypothesis from the transition destination dialog state determination operation determination unit into one, and confirms the recognition result to the user to determine the transition destination dialog state hypothesis. The dialogue state is determined and output, and the stored transitional dialogue state hypothesis is rewritten in the dialogue state transition storage unit. Of the destination dialog state In this case, the transition destination dialog state hypothesis is input, the provisional transition destination dialog state is determined and output, and the transition destination dialog state hypothesis is rewritten to the dialog state transition storage unit. The dialog state transition history from the start of the dialog and the transition destination dialog state determination unit or the transition destination dialog state hypothesis from the provisional transition destination dialog state determination unit are stored, and the dialog operation execution unit stores the transition destination dialog state determination unit or The transition destination dialog state from the provisional transition destination dialog state determination unit is input, a system response defined in the transition destination dialog state is output, and the recognition target vocabulary defined in the transition destination dialog state is recognized by the speech recognition unit. 2. The speech recognition unit according to claim 1, wherein the speech recognition unit performs speech recognition on the input speech using a recognition target vocabulary input from the interactive operation execution unit, and outputs a plurality of recognition results. sound of Voice interaction device. 3. The speech recognition unit is configured to output a plurality of recognition results and a score of the recognition results, and the transition destination dialog state determination operation determining unit determines a score of the recognition result input from the speech recognition unit. The voice interaction apparatus according to claim 1, wherein whether to perform a confirmation operation is determined according to the condition. 4. Each of the dialog states stored in the dialog procedure storage unit describes whether or not it is necessary to perform a definite operation in order to perform a state transition from another dialog state to the dialog state. The destination dialog state determining operation determining unit determines that the hypothesis of the destination dialog state determined from the recognition result input from the voice recognition unit, the contents of the dialog state transition storage unit, and the dialog procedure requires that the determining operation be performed in advance. The voice interaction device according to any one of claims 1 to 3, wherein it is determined that a confirmation operation is performed. 5. The transition destination dialog state determining operation determining unit determines a recognition result from the voice recognition unit to uniquely determine an item value for an uninput item even if all input items from the user have not been input. The voice interaction apparatus according to claim 1, wherein when it is determined to be, a decision operation is performed. 6. The transition destination dialog state determining operation determining unit determines whether or not to perform a determining operation according to a system response defined in the transition destination dialog state hypothesis.
6. The voice interaction device according to any one of items 5 to 5. 7. A transition destination dialog state deciding operation determining unit decides that a decision operation is to be performed when there is no common system response in the transition destination dialog state hypothesis, and a common system response exists in the transition destination dialog state hypothesis. 7. The voice interaction apparatus according to claim 6, wherein, when performing, only the transition destination dialog state hypothesis having a common system utterance is output as the transition destination dialog state hypothesis. 8. A system in which a plurality of system responses can be described in each dialog state stored in the dialog procedure storage unit, and the dialog operation execution unit is configured to input a transition destination dialog state from the temporary transition destination dialog state determination unit. Outputting a system response defined in the input destination dialog state that is common to the system response specified in the destination dialog state hypothesis stored in the interaction state transition storage unit, among the system responses specified in the input destination state. Item 7. The voice interaction device according to item 5 or 6. 9. The transition destination dialog state determining operation determining unit determines that the determining operation is performed when the total vocabulary of all the recognition target vocabularies of the transition destination dialog state hypothesis is larger than a predetermined reference. The speech dialogue device according to claim 1, wherein: 10. The transition destination dialog state determining operation determining unit refers to the dialog state transition storage unit and determines that the length of the transition sequence from the determined dialog state to the transition destination dialog state hypothesis is not less than a predetermined reference value. The voice interaction device according to claim 1, wherein it is determined that a decision operation is performed in such a case.