JP6495015B2 - Spoken dialogue control device, control method of spoken dialogue control device, and spoken dialogue device - Google Patents

Description

  The present invention relates to a voice dialogue control device for controlling a voice dialogue device that responds to a user's utterance.

  2. Description of the Related Art Conventionally, a speech dialogue apparatus (robot) that interacts with a user by responding to the user's utterance with voice or motion has been widely studied. Here, in the dialogue between the user and the voice interactive device, a certain amount of time is required from when the user speaks until the voice interactive device responds according to the content of the speech. If the voice interaction device does not operate at this time, the user may feel stress in communication with the voice interaction device. As a solution to this problem, for example, in Patent Document 1 below, a standby time until receiving an answer from the server is predicted, and if necessary, an operation (filling operation) is performed to fill the standby time. Technology is disclosed.

JP 2014-191030 A (released on October 06, 2014) JP 2003-330923 A (published November 21, 2003)

  However, with the technologies described in Patent Documents 1 and 2, the response when the user utters the same content is uniform regardless of whether or not the joining operation is performed. For example, when you respond immediately to the user's question "What is the weather today?" Is executing the response. In other words, the conventional technology has a problem that the voice interaction apparatus can perform only a uniform response in a situation where a response different from the normal response is required, for example, it takes time to respond.

  The present invention has been made in view of the above problems, and an object of the present invention is to improve the flexibility of communication between a user and a voice interaction device by correcting the response in a situation where the response should be corrected. It is to provide a control device and the like.

  In order to solve the above-described problem, a voice interaction control device according to one aspect of the present invention acquires a response, a response generation unit that generates a response to a voice uttered by a user, and that is executed by the voice interaction device. The determination unit for determining whether or not the correction condition for determining whether or not the response needs to be corrected is satisfied during the standby time until the response can be output, and the determination unit satisfies the correction condition. A correction unit that generates a correction response that corrects the response generated by the response generation unit, and a response execution unit that causes the voice interaction device to execute the correction response generated by the correction unit. Prepare.

  In order to solve the above-described problem, a control method for a voice interaction control device according to an aspect of the present invention includes a response generation step of generating a response to a voice uttered by a user, executed by the voice interaction device; A determination step for determining whether or not a correction condition for determining whether or not the response needs to be corrected is satisfied in a waiting time from when the voice is acquired until the response can be output. When it is determined that the correction condition is satisfied, a correction step for correcting the response generated in the response generation step is generated, and the correction response generated in the correction step is sent to the voice interaction device. A response execution step to be executed.

  According to one aspect of the present invention, there is an effect that the flexibility of communication between the user and the voice interaction apparatus is improved.

It is a block diagram which shows the structure of the voice interactive apparatus which concerns on Embodiment 1 of this invention. It is a figure which shows the data structure and data example of a response correction table which are memorize | stored in the memory | storage part of the voice interactive apparatus shown in FIG. It is a flowchart which shows an example of the flow of the response execution process which the voice dialogue control apparatus shown in FIG. 1 performs. It is a block diagram which shows the structure of the voice interactive apparatus which concerns on Embodiment 2 of this invention. It is a figure which shows the data structure and example of a data of the calling attribute table memorize | stored in the memory | storage part of the voice interactive apparatus shown in FIG. It is a figure which shows the data structure and data example of a response correction table which are memorize | stored in the memory | storage part of the voice interactive apparatus shown in FIG. It is a flowchart which shows an example of the flow of the response execution process which the voice dialogue control apparatus shown in FIG. 4 performs. It is a block diagram which shows the structure of the voice interactive apparatus which concerns on Embodiment 3 of this invention. FIG. 9 is a diagram illustrating a data structure and a data example of a connection operation table stored in a storage unit of the voice interactive apparatus illustrated in FIG. 8. It is a flowchart which shows an example of the flow of the response execution process which the voice dialogue control apparatus shown in FIG. 8 performs. It is a flowchart which shows an example of the flow of the joining operation | movement determination processing in the flowchart shown in FIG.

Embodiment 1
An embodiment (Embodiment 1) of the present invention will be described below with reference to FIGS.

  First, the voice interactive apparatus 10 according to the present embodiment will be described with reference to FIG. FIG. 1 is a block diagram showing a configuration of a voice interaction apparatus 10 according to the present embodiment.

  The voice interaction device 10 is a device that interacts with a user by responding to the user's utterance with a voice or action. In addition, although a humanoid robot is mentioned as a specific example of the voice interactive apparatus 10, it is not limited to this. For example, other specific examples of the voice interaction device 10 include a mobile terminal with a voice interaction function such as a smartphone, a car navigation system with a voice interaction function, and the like. As shown in FIG. 1, the voice interaction device 10 includes a voice interaction control device as the control unit 1. Note that the voice interaction device 10 and the voice interaction control device may be separate. The voice interaction device 10 includes a voice input unit 2, a communication unit 3, a voice output unit 4, a drive unit 5, and a storage unit 6 in addition to the control unit 1 (voice conversation control device).

  The voice input unit 2 is a so-called microphone that acquires voice uttered by the user. The voice input unit 2 converts the acquired voice into voice data and outputs the voice data to the voice recognition unit 13 described later (indicated by d2 in FIG. 1). Further, the voice input unit 2 notifies the standby time measuring unit 11 described later that voice has been acquired (indicated by d1 in FIG. 1). In the communication unit 3, the voice interaction device 10 communicates with an external device. Specifically, the communication unit 3 is controlled by a response generation unit 14 described later, and receives data necessary for generating a response from an external device. For example, the communication unit 3 acquires data on tomorrow's weather from a weather forecast server (not shown) that manages data on weather forecasts, and outputs the data to the response generation unit 14. The audio output unit 4 is a so-called speaker that outputs audio. Specifically, the voice output unit 4 outputs a voice as a response to the voice uttered by the user. The drive unit 5 drives a movable part such as a head or a leg in the voice interaction device 10 (humanoid robot), and is a servo motor, for example. An actuator other than the servo motor may be used. Specifically, the drive unit 5 causes the voice interaction apparatus 10 to perform an operation as a response to the voice uttered by the user by driving the movable part. In addition, when the voice interaction apparatus 10 is an apparatus that does not have a movable part such as a smartphone, the driving unit 5 may be omitted. The storage unit 6 stores various data used in the voice interaction device 10. The storage unit 6 stores at least a response correction table 61. Details of the response correction table 61 will be described later.

  The control unit 1 performs overall control of each unit included in the voice interaction device 10. The control unit 1 includes a standby time measurement unit 11, a response correction unit 12, a voice recognition unit 13, a response generation unit 14, and a response execution unit 15.

The standby time measuring unit 11 measures a standby time from when a voice uttered by the user is acquired until a response to the voice is generated. Specifically, when the standby time measuring unit 11 is notified from the voice input unit 2 that voice has been acquired, the standby time measuring unit 11 starts measuring time using a timer (not shown). In addition, when the standby time measuring unit 11 receives notification from the response generating unit 14 that the generation of response information has been completed, the standby time measuring unit 11 ends the time measurement by the timer. Then, it outputs the measured time T _a to the response modification unit 12.

  The response correction unit 12 corrects the response information generated by the response generation unit 14 to generate a correction response. The response correction unit 12 includes a correction necessity determination unit 21 (determination unit) and a correction execution unit 22 (correction unit).

  Whether or not the correction necessity determination unit 21 acquires a voice uttered by the user and whether or not a correction condition for determining whether or not the response needs to be corrected is satisfied in a waiting time until a response to the voice can be output. Determine whether. Specifically, the correction necessity determination unit 21 determines whether or not the time received from the standby time measurement unit 11 (the time measured by the standby time measurement unit 11) is greater than a predetermined value. Then, the determination result is output to the correction execution unit 22. The predetermined value is a time value (for example, 3 seconds) at which it is determined that the generation of response information has been prolonged.

  The correction execution unit 22 corrects the response information generated by the response generation unit 14 when it is determined that the correction condition is satisfied. Specifically, the correction execution unit 22 stores the determination result received from the correction necessity determination unit 21 in the storage unit 6 when the time received from the standby time measurement unit 11 is greater than a predetermined value. The response information received from the response generation unit 14 is corrected using the stored response correction table 61.

  Here, the details of the response correction table 61 will be described with reference to FIG. FIG. 2 is a diagram illustrating a data structure and a data example of the response correction table 61 stored in the storage unit 6. The response correction table 61 shown in FIG. 2 is an example, and the data structure and data example are not limited to the example in FIG. The response correction table 61 includes information indicating time required for generating response information, that is, information indicating standby time (time information, hereinafter referred to as standby time), and information indicating content of additional response (correction content) added to the response information. It is a table in which (correction content information, hereinafter, additional response information) is associated. That is, the response correction table 61 is a table in which different additional response information is associated with the time measured by the standby time measuring unit 11. The “standby time” column may store information indicating a time range such as “4 to 7 seconds”.

  More specifically, the correction execution unit 22 refers to the response correction table 61 and identifies additional response information associated with a standby time that matches (corresponds to) the time received from the standby time measurement unit 11. . Then, the response information is corrected by reading the specified additional response information and adding it to the received response information. For example, when response information (voice data) indicating a response to utter “It's sunny” is received, and the received time is 5 seconds, the correction execution unit 22 displays No. 2 shown in FIG. 1 or No. One of the two additional response information is determined as additional response information for correcting the response information. As described above, when there are a plurality of additional response information that matches the associated waiting time, the correction execution unit 22 may select one at random. In this case, no. It is assumed that 1 additional response information is selected. The correction execution unit 22 corrects the voice data for uttering “It's sunny” to voice data uttering “Please wait. It ’s sunny”. Then, the correction execution unit 22 outputs the corrected response information (speech data in the above example) to the response execution unit 15. Note that the received time and the time stored in the “waiting time” column may not completely match, and the received time is included in a predetermined range including the stored waiting time. In addition, additional response information associated with the waiting time may be added to the response information.

  On the other hand, the correction execution unit 22 does not correct the response information received from the response generation unit 14 when the determination result received from the correction necessity determination unit 21 indicates that the standby time is equal to or less than a predetermined value. Then, it is output to the response execution unit 15 as it is.

  The voice recognition unit 13 performs voice recognition processing on the voice data received from the voice input unit 2. Note that existing technology can be used for the speech recognition processing. The voice recognition unit 13 outputs the voice recognition result of the received voice data to the response generation unit 14.

  The response generation unit 14 generates response information indicating a response to the voice uttered by the user. There are three types of responses: voice output, movement of the moving part of the voice interaction device 10, and voice output and movement of the movable part. An existing technique can be used to generate response information by the response generation unit 14. For example, a table (not shown) in which the content of the voice data recognized and the response content are associated with each other may be stored in the storage unit 6, and the response information may be generated by referring to the table. Further, when it is necessary to use external data such as tomorrow's weather information to generate response information, the response generation unit 14 controls the communication unit 3 to acquire the data. The response generation unit 14 outputs the generated response information (voice data for voice output, action data for operating the movable part, etc.) to the response correction unit 12 (correction execution unit 22) (d4 in FIG. 1). ). In addition, the response generation unit 14 notifies the standby time measurement unit 11 that the generation of response information has been completed (indicated by d3 in FIG. 1).

  The response execution unit 15 executes a response according to the response information generated by the response generation unit 14 and corrected by the response correction unit 12 as necessary. Specifically, the response execution unit 15 receives response information from the response correction unit 12 (correction execution unit 22), and causes the voice interaction device 10 to execute the operation indicated by the response information. For example, the voice output unit 4 is controlled to output a voice, or the drive unit 5 is controlled to operate a movable part of the voice interaction device 10.

  Next, based on FIG. 3, the flow of response execution processing executed by the control unit 1 will be described. FIG. 3 is a flowchart illustrating an example of a response execution process executed by the control unit 1.

  First, the voice input unit 2 waits for voice input (S1). When the voice input unit 2 acquires the voice uttered by the user (YES in S1), the voice input unit 2 converts the acquired voice into voice data, and outputs the voice data to the voice recognition unit 13. In addition, the voice input unit 2 notifies the standby time measurement unit 11 that voice has been acquired.

Subsequently, when receiving the notification that the voice has been acquired, the standby time measuring unit 11 starts measuring time (S2). The voice recognition unit 13 performs voice recognition processing on the received voice data (S3). The voice recognition unit 13 outputs the voice recognition result to the response generation unit 14. Subsequently, the response generation unit 14 generates response information according to the received voice recognition result (S4, response generation step). Then, the response generation unit 14 notifies the standby time measurement unit 11 that the response information has been generated. In addition, the response generation unit 14 outputs the generated response information to the correction execution unit 22. Subsequently, when receiving the notification from the response generation unit 14, the standby time measurement unit 11 ends the time measurement (S5). Then, the measurement result (measured time T _a ) is output to the correction necessity determination unit 21.

The correction necessity determination unit 21 determines whether or not the time Ta received from the standby time measurement unit 11 is greater than _a predetermined value (S6, determination step). Then, the correction necessity determination unit 21 outputs the determination result to the correction execution unit 22.

If the time T _a is determined to be greater than the predetermined value (YES in S6), the correction execution unit 22 receives the determination result, to identify additional responses as a function of time T _a, modifies the response information ( S7, correction step). Specifically, correction execution unit 22 refers to the response modification table 61 stored in the storage unit 6 to identify the additional response information associated with the waiting time that matches the received time T _a. Then, the identified additional response information is read out and added to the response information received from the response generation unit 14, thereby correcting the response information. Then, the correction execution unit 22 outputs the corrected response information to the response execution unit 15.

In contrast, if the time T _a is determined to be equal to or less than the predetermined value (NO at S6), correction execution unit 22 receives the determination result, corrects the response information received from the response generation unit 14 Without being output to the response execution unit 15. That is, the process in step S7 described above is omitted.

  Finally, the response execution unit 15 causes the voice interaction device 10 to execute a response (S8, response execution step). Specifically, the response execution unit 15 controls the voice output unit 4 to output voice according to the received response information, or controls the drive unit 5 to operate the movable part of the voice interaction device 10. Or Thus, the response execution process ends.

[Embodiment 2]
The following will describe another embodiment (Embodiment 2) of the present invention with reference to FIGS. For convenience of explanation, members having the same functions as those described in the embodiment are given the same reference numerals, and descriptions thereof are omitted.

  In the present embodiment, the voice input from the voice input unit 2 before identifying the additional response information in which the time received from the standby time measuring unit 11 matches the standby time stored in the “standby time” column ( A description will be given of an example of identifying additional response information in which the attribute of (sound data) matches the attribute of additional response information.

  First, the voice interactive apparatus 10a according to the present embodiment will be described with reference to FIG. FIG. 4 is a block diagram showing the configuration of the voice interaction apparatus 10a according to this embodiment. Compared with the voice interaction device 10 according to the first embodiment, the voice interaction device 10 a includes a control unit 1 a instead of the control unit 1 and includes a storage unit 6 a instead of the storage unit 6. In this embodiment, compared with the control unit 1 according to the first embodiment, the control unit 1a includes a response correction unit 12a instead of the response correction unit 12, and replaces the voice recognition unit 13 with a voice recognition unit 13a (voice Attribute specific part). Further, the storage unit 6 a stores a response correction table 61 a instead of the response correction table 61 as compared with the storage unit 6 according to the first embodiment. In addition, the storage unit 6a newly stores a call attribute table 62.

  The voice recognition unit 13a performs voice recognition processing on the voice data received from the voice input unit 2, and outputs a voice recognition result to the response generation unit 14 (indicated by d5 in FIG. 4). Further, the voice recognition unit 13a gives an attribute to the voice data. Specifically, the voice recognition unit 13a refers to the call attribute table 62 stored in the storage unit 6a after performing voice recognition.

  Here, the details of the call attribute table 62 will be described with reference to FIG. FIG. 5 is a diagram showing a data structure and a data example of a call attribute table stored in the storage unit 6a. The call attribute table 62 shown in FIG. 5 is an example, and the data structure and data example are not limited to the example of FIG. The call attribute table 62 is a table in which a speech recognition result and a call attribute are associated with each other as shown in FIG. In other words, the call attribute table 62 is a table for specifying call attributes of voice data based on the voice recognition result. In the “calling” column, a speech recognition result, that is, information obtained by converting the speech uttered by the user into text data is stored. In the “calling attribute” column, a calling attribute (sound attribute) indicating a category of the content of the sound is stored. The voice recognition unit 13a refers to the call attribute table 62 using the voice recognition result, and specifies the call attribute of the voice data. Then, the specified call attribute is output to the correction execution unit 22a (indicated by d6 in FIG. 4). For example, when the voice data is “What is the weather today?” As a result of the voice recognition, the voice recognition unit 13a sets “question” and “weather” as the calling attributes of the voice data. Output to.

  The call attribute table 62 may be as shown in FIG. That is, a keyword included in the voice uttered by the user is stored in the “calling” column, and a calling attribute may be associated with each keyword. In this case, the voice recognition unit 13a specifies all the call attributes associated with the keywords included in the voice data, and outputs the specified call attributes to the correction execution unit 22a.

  The response correction unit 12a corrects the response information generated by the response generation unit 14. The response correction unit 12a includes a correction necessity determination unit 21 and a correction execution unit 22a. Since the correction necessity determination unit 21 has already been described in the first embodiment, the description thereof is omitted here.

  The correction execution unit 22a corrects the response information generated by the response generation unit 14. Specifically, when the determination result received from the correction necessity determination unit 21 indicates that the time received from the standby time measurement unit 11 is greater than a predetermined value, the correction execution unit 22a stores the storage unit 6 in the storage unit 6. The response is corrected using the stored response correction table 61a.

  Here, the details of the response correction table 61a will be described with reference to FIG. FIG. 6 is a diagram illustrating a data structure and a data example of the response correction table 61a stored in the storage unit 6a. In the response correction table 61a, an additional response attribute indicating the category of the content of the additional response indicated by the additional response information is further associated with the standby time and the additional response information. The “additional response attribute” column stores the additional response attribute described above. The additional response attribute “question” indicates that the additional response indicated by the additional response information associated with the additional response attribute is suitable for addition to a response indicating an answer to the question. The additional response attribute “all” indicates that the additional response indicated by the additional response information associated with the additional response attribute is suitable for addition to all responses regardless of the content of the response.

  More specifically, when the determination result received from the correction necessity determination unit 21 indicates that the time received from the standby time measurement unit 11 is greater than a predetermined value, the correction execution unit 22a first performs voice An additional response attribute that matches the call attribute received from the recognition unit 13a is specified from the response correction table 61a. When there is an additional response attribute that matches the call attribute in addition to “all”, the correction execution unit 22a specifies only the additional response information associated with the additional response attribute. Then, the correction execution unit 22a identifies the additional response information associated with the time received from the standby time measuring unit 11 from the additional response information associated with the identified additional response attribute. Then, the response information is corrected by reading the specified additional response information and adding it to the received response information.

  For example, response information (voice data) indicating a response to utter “It's sunny” is received from the response generation unit 14, and call attributes “question” and “weather” are received from the voice recognition unit 13 a, When the received time is 5 seconds, the correction execution unit 22a firstly selects a No. that matches the received call attribute “question”. 1 and no. 3 additional response information is specified. Then, the received time matches the response generation processing time associated with the additional response information. One additional response information is determined as additional response information for correcting the response information. Then, the correction execution unit 22a corrects the voice data for uttering "It's sunny" to voice data uttering "Please wait. It's sunny". In the above-described example, when a plurality of call attributes are received, the additional response information associated with the additional response attribute that matches at least one of the call attributes is specified. However, the present invention is not limited to this. Alternatively, additional response information associated with additional response attributes that completely match a plurality of call attributes may be specified.

  Next, the flow of response execution processing executed by the control unit 1a will be described with reference to FIG. FIG. 7 is a flowchart showing an example of a response execution process executed by the control unit 1a. Here, only differences from the response execution process described in the first embodiment will be described. Specifically, Step S11 to Step S13, Step S15 to Step S17, and Step S20 are the same as Step S1 to Step S3, Step S4 to Step S6, and Step S8 in the flowchart of FIG. The description in is omitted.

  After performing the voice recognition process, the voice recognition unit 13a specifies the call attribute of the recognized voice (S14). Specifically, the voice recognition unit 13a refers to the call attribute table 62 using the voice recognition result, and specifies the call attribute of the voice data. Then, the specified call attribute is output to the correction execution unit 22a.

If the time T _a is determined to be greater than the predetermined value (YES in S17), the correction execution unit 22 receives the determination result, correspondence between the additional response attribute that matches the call attributes received from the voice recognition unit 13a The added additional response information is specified (S18). Specifically, the correction execution unit 22 refers to the response correction table 61a stored in the storage unit 6 and specifies additional response information associated with the additional response attribute that matches the call attribute. Then, from among the identified additional response information, and further identifies the additional response information corresponding to the time T _a, Correct Answers (S19). Then, the correction execution unit 22 outputs the corrected response information to the response execution unit 15.

Incidentally, if the time T _a is determined to be equal to or less than the predetermined value (NO in S17), the correction execution unit 22 receives the determination result, without modifying the response information received from the response generation unit 14, The response is output to the response execution unit 15. That is, the processes of step S18 and step S19 are omitted.

  In the present embodiment, the example has been described in which the voice recognition unit 13a specifies the call attribute of the voice data received from the voice input unit 2. On the other hand, the response generation unit 14a (not shown) may specify the response attribute of the generated response information. Specifically, a response attribute table 62a (not shown), which is a table in which response information and response attributes are associated with each other, is stored in the storage unit 6a in place of the call attribute table 62. The response generation unit 14a The response attribute is specified from the response attribute table 62a using the generated response information. Then, the generated response information is associated with the identified response attribute and output to the correction execution unit 22a.

[Embodiment 3]
The following will describe still another embodiment (Embodiment 3) of the present invention with reference to FIGS. For convenience of explanation, members having the same functions as those described in the embodiment are given the same reference numerals, and descriptions thereof are omitted.

  In the first and second embodiments described above, it is determined whether or not the response is to be corrected by determining whether or not the time measured by the standby time measuring unit 11 is greater than a predetermined value. On the other hand, in the present embodiment, during the standby time, that is, the time from when the voice uttered by the user is acquired to when the response to the voice is generated, the voice interaction apparatus performs at least one of voice output and operation. An example will be described in which it is determined whether or not the response is to be corrected by determining whether or not it has been determined (or whether or not it has actually been performed).

  First, the voice interactive apparatus 10b according to the present embodiment will be described with reference to FIG. FIG. 8 is a block diagram showing the configuration of the voice interaction apparatus 10b according to the present embodiment. Compared with the voice interaction device 10 according to the first embodiment, the voice interaction device 10 b includes a control unit 1 b instead of the control unit 1, a voice input unit 2 b instead of the voice input unit 2, and a storage unit 6. Instead, a storage unit 6b is provided. In the present embodiment, the control unit 1b does not include the standby time measurement unit 11 as compared with the control unit 1 according to the first embodiment. In addition, a response correction unit 12b is included instead of the response correction unit 12. Further, a standby time prediction unit 16 and a field connection operation control unit 17 are newly included. In addition, the storage unit 6b stores a new connection operation table 63 in comparison with the storage unit 6 according to the first embodiment.

  The voice input unit 2b converts the acquired voice into voice data and outputs the voice data to the voice recognition unit 13 described later (indicated by d8 in FIG. 8). The voice input unit 2 outputs at least one of the size (data amount) and time (speech time) of the voice data (hereinafter referred to as voice attached information) to the standby time prediction unit 16 (in FIG. 8, at d7). Shown).

Waiting time prediction unit 16 predicts the acquired voice voice dialogue system 10b has issued the user, the waiting time T _b to response to the voice can be outputted. Specifically, first, when the standby time predicting unit 16 receives the audio attached information from the audio input unit 2, the standby time predicting unit 16 predicts the standby time using the size (data amount) of the audio data. More specifically, the standby time prediction unit 16 uses a calculation formula “standby time T _b = α × data amount (α is a standby time required per unit data amount, which is a predetermined value)”. The estimated waiting time _Tb is calculated. The standby time prediction unit 16 outputs the predicted (calculated) predicted standby time _Tb to the later-described place-joining operation determination unit 71 and the correction execution unit 22b. Note that the standby time prediction unit 16 may predict the standby time using the time of the voice data (user's speech time). Specifically, the waiting time prediction unit 16 uses a calculation formula “waiting time T _b = β × speech time (β is a waiting time required per unit utterance time, which is a predetermined value)”. it may calculate the predicted waiting time T _b. Further, the standby time may be predicted (calculated) using both the data amount of the voice data and the speech time. When the predicted standby time calculated from the data amount is different from the predicted standby time calculated from the utterance time, the longer (or shorter) predicted standby time may be adopted, and an average value of the two predicted standby times may be calculated. It may be calculated and the calculated average waiting time may be output to the connection determination unit 71.

  The field connection operation control unit 17 determines and executes the field connection operation. The field connection operation control unit 17 includes a field connection operation determination unit 71 and a field connection operation execution unit 72.

If connecting operation determining unit 71, based on the standby time predicting section predicts the waiting time 16 predicts T _b, is what determines the field joint operation voice dialogue system 10 performs. Here, the place-joining operation is an operation that is executed by the voice interaction apparatus 10 during a time (waiting time) from when a voice uttered by the user is acquired until a response to the voice can be output. Specifically, the field joint operation determining unit 71, by using a field joint operation table 63 stored in the storage unit 6, and the predicted waiting time predicted waiting time prediction unit 16 T _b, the sound during the waiting time The place-joining operation is determined according to the place-joining operation time required for the place-joining operation to be executed by the interactive apparatus 10b.

  Here, with reference to FIG. 9, the details of the joining operation table 63 will be described. FIG. 9 is a diagram illustrating a data structure and a data example of the joining operation table 63 stored in the storage unit 6b. 9 is merely an example, and the data structure and the data example are not limited to the example in FIG. The field connection operation table 63 is a table in which information indicating a field connection operation is associated with a field connection operation time which is a time required for the field connection operation. Stored in the “place-joining action” column is information on a plurality of action candidates (hereinafter referred to as place-joining action information) indicating actions that can be executed by the voice interaction apparatus 10. In the “type” column, whether the connection operation at each place is to output a voice (indicated by “voice” in FIG. 9), or to move the movable part of the voice interactive device 10b ( Information indicating whether or not to perform both (indicated by “gesture” in FIG. 9) (indicated by “voice + gesture” in FIG. 9) is stored. In the column of “joining operation time”, the above-mentioned joining operation time is stored.

More specifically, the field joint operation determining unit 71, the predicted waiting time T _b received, by subtracting the respective field connecting operation time field joint operation table 63, a subtraction value T _c of each spot joining operation information calculate. In addition, c is a number stored in “No.” in the field connection operation table 63. Subsequently, the field connection operation determination unit 71 sets the calculated subtraction value T _c to 0 or more, and after the voice interaction apparatus 10b executes the field connection operation until the generation of the response is completed. It is determined whether or not 10b is equal to or shorter than a first allowable time X indicating a time that is allowable as a time during which no operation is performed (whether there is connection operation information that satisfies 0 ≦ T _c ≦ X). The first allowable time X is a preset value. For example, if X = 2, the allowable time from the completion of the splicing operation until the generation of the response is 2 seconds. That is.

When there is field connection operation information satisfying 0 ≦ T _c ≦ X, the field connection operation determination unit 71 determines the field connection operation indicated by the field connection operation information as a field connection operation to be executed by the voice interaction apparatus 10b, and The field connection operation information is output to the field connection operation execution unit 72. For example, when the predicted standby time _Tb is 2 seconds and the first allowable time X = 1, No. 1 shown in FIG. 2 and no. The field connection operation information of 3 satisfies 0 ≦ T _c ≦ X. Therefore, the spot-linking operation determining unit 71 determines the No. 2 or No. 3 is read out and output to the connecting operation executing unit 72.

In addition, when there are a plurality of field connection operation information satisfying 0 ≦ T _c ≦ X, the field connection operation information with a smaller value of the subtraction value T _c is used in order to shorten the time during which the voice interaction device 10b does not execute the operation. Is preferably selected. In other words, in the case of the above example, subtraction value _{T c} is 0 No. It is preferable to select the operation information for connecting the three. In addition, when there are a plurality of field connection operation information having the same value of the subtraction value _Tc , one of the field connection operation information may be selected at random.

On the other hand, when there is no field joining operation information that satisfies 0 ≦ T _c ≦ X, the field joining operation determining unit 71 sets 0 for each of the sign change values −T _c that are values obtained by changing the sign of the subtraction value T _c. Whether or not it is equal to or shorter than the second permissible time Y indicating the time that is acceptable as the time from the completion of the response generation to the completion of the connection operation of the voice interactive device 10b (0 ≦ −T _c ≦ Y) It is determined whether or not there is connection operation information that satisfies the above condition. The second allowable time Y is a preset value. For example, if Y = 2, the allowable time from the completion of response generation to the completion of the splicing operation is 2 seconds. That is. Incidentally, the field joint operation determining unit 71 may calculate the code conversion value -T _c by subtracting the predicted waiting time T _b which is received from the field connecting operation time.

When there is field connection operation information satisfying 0 ≦ −T _c ≦ Y, the field connection operation determination unit 71 determines a field connection operation indicated by the field connection operation information as a field connection operation to be executed by the voice interaction apparatus 10b. The field connection operation information is output to the field connection operation execution unit 72. For example, when the predicted waiting time _Tb is 1 second and the second allowable time Y = 1, No. 1 shown in FIG. 2 and no. The field connection operation information of 3 satisfies 0 ≦ −T _c ≦ Y. For this reason, the field connection operation determination unit 71 determines whether the No. 2 or No. 3 is read out and output to the connecting operation executing unit 72.

In addition, when there are a plurality of field connection operation information satisfying 0 ≦ −T _c ≦ Y, a field connection operation with a smaller code conversion value −T _c is performed in order to shorten the time during which the voice interactive device 10b does not execute the operation. It is preferable to select information. In other words, in the case of the above example, the code conversion value −T _c becomes 0. It is preferable to select the operation information for connecting the two points.

Also, if the value of the code conversion value -T _c there are multiple same field joint operation information, a field joint operation information may be selected in a single random from those.

  Note that, in at least one of the first allowable time X and the second allowable time Y, the same value may be set for all the joining motion information, or different values may be set for each joining motion information. Good. Further, at least one of the first allowable time X and the second allowable time Y may be set according to at least one of the amount of audio data and the speech time. In other words, the field connection operation determination unit 71 determines at least one of the first allowable time X and the second allowable time Y based on the data amount or speech time of the audio data received from the standby time prediction unit 16.

On the other hand, when there is no field connection operation information that satisfies 0 ≦ −T _c ≦ Y, the field connection operation determination unit 71 selects a plurality of field connection operation information. Specifically, the field connection operation determination unit 71 selects one field connection operation information having the longest field connection operation time among the field connection operation information satisfying the field connection operation time ≦ the standby time. Then, from the predicted waiting time T _b, calculates a subtracted value (remaining time) the field connecting operation time associated with the selected field joint operation information, the field joint operation information satisfying a field joint operation time ≦ remaining time Select further. Then, the field connection operation determination unit 71 calculates a total value obtained by summing the field connection operation times associated with the selected plurality of field connection operation information, and 0 ≦ predicted standby time T _b −total value ≦ X or 0 It is determined whether or not ≦ − (predicted waiting time T _b −total value) ≦ Y is satisfied. If either one is satisfied, a plurality of field connection operation information is output to the field connection operation execution unit 72 in association with the numbers in the “No.” column.

On the other hand, it is not satisfied either, calculates a value obtained by subtracting the sum from the prediction waiting time T _b, further selects a field joint operation information satisfying the value calculated field joint operation time ≦ the. Then, the field connection operation determination unit 71 calculates a total value obtained by summing the field connection operation times associated with the selected plurality of field connection operation information, and 0 ≦ predicted standby time T _b −total value ≦ X or 0 It is determined whether or not ≦ − (predicted waiting time T _b −total value) ≦ Y is satisfied. The field connection operation determination unit 71 performs these processes until either 0 ≦ predicted standby time T _b −total value ≦ X or 0 ≦ − (predicted standby time T _b −total value) ≦ Y is satisfied. repeat.

In addition, when there are a plurality of connection operation information satisfying 0 ≦ T _c ≦ X or 0 ≦ −T _c ≦ Y, in order to shorten the time during which the voice interactive apparatus 10b does not execute an operation, T _c or −T It is preferable to select the joining operation information when the value of _c is smaller. Incidentally, if the value of T _c or -T _c there are multiple same field joint operation information, a field joint operation information may be selected in a single random from those.

  The field connection operation execution unit 72 causes the voice interaction device 10b to execute the field connection operation indicated by the field connection operation information determined by the field connection operation determination unit 71. Specifically, when the field connection operation execution unit 72 receives the field connection operation information from the field connection operation determination unit 71, the field connection operation execution unit 72 causes the voice interaction device 10b to execute the field connection operation indicated by the field connection operation information. For example, the voice output unit 4 is controlled to output a voice, or the drive unit 5 is controlled to operate a movable part of the voice interaction device 10b. When the execution of the joining operation is completed, the jumping operation execution unit 72 notifies the response execution unit 15 to that effect. In addition, when receiving a plurality of field connection operation information from the field connection operation determination unit 71, the field connection operation execution unit 72 randomly determines an operation order of the plurality of field connection operations. Then, the voice interaction device 10 is caused to execute the joining operation in the determined operation order. Note that information defining the operation order may be stored in the storage unit 6b, and the operation order may be determined based on the operation order indicated by the information, or the operation order may be associated with the jumping operation. “No.” may be in ascending order. It should be noted that the field connection operation control unit 17 outputs the field connection operation information indicating the field connection operation to be executed by the voice interaction apparatus 10b to the correction necessity determination unit 21b. This may be performed at the time when the field connection operation determining unit 71 determines the field connection operation information indicating the field connection operation to be executed by the voice interaction device 10b, or the field connection operation execution unit 72 is connected to the voice interaction device 10b. You may perform after performing a joining operation | movement. In the flowchart to be described later, it is assumed that the joining operation execution unit 72 causes the voice interaction device 10b to perform the joining operation, and then outputs the joining operation information to the correction necessity determination unit 21b.

  The response correction unit 12b corrects the response information generated by the response generation unit 14. The response correction unit 12b includes a correction necessity determination unit 21b and a correction execution unit 22b.

  Whether or not the correction necessity determination unit 21b has acquired the voice uttered by the user and whether or not the correction condition for determining whether or not the response needs to be corrected is satisfied during the waiting time until the response to the voice can be output is satisfied. Determine whether. Specifically, the correction necessity determination unit 21b indicates a field connection operation in which the field connection operation information received from the field connection operation determination unit 71 causes the voice interaction apparatus 10b to perform at least one of voice output and operation. It is determined whether or not. More specifically, the correction necessity determination unit 21b determines that the received connection operation information among the connection operation information illustrated in FIG. It is determined whether the operation information is one field connection operation information or the other field connection operation information. Then, the determination result is output to the correction execution unit 22b.

When it is determined that the correction condition is satisfied, the correction execution unit 22b corrects the response information generated by the response generation unit 14. Specifically, the correction execution unit 22b indicates that the determination result received from the correction necessity determination unit 21b indicates that the connection operation information received by the correction necessity determination unit 21b is No. When the identification information indicates that the field joint operation information other than 1, with reference to the response modification table 61, specifies the additional response information corresponding to the predicted waiting time T _b which is received from the standby time predicting unit 16. Then, the response information is corrected by reading the specified additional response information and adding it to the received response information. Then, the correction execution unit 22b outputs the corrected response information to the response execution unit 15.

  On the other hand, the correction execution unit 22b indicates that the determination result received from the correction necessity determination unit 21b indicates that the connection operation information received by the correction necessity determination unit 21b is No. When it is shown that the operation information is connected to the field 1, the response information received from the response generation unit 14 is output to the response execution unit 15 without modification.

  Next, based on FIG. 10, the flow of response execution processing executed by the control unit 1b will be described. FIG. 10 is a flowchart illustrating an example of a flow of response execution processing executed by the control unit 1b.

  First, the voice input unit 2 waits for voice input (S21). When the voice input unit 2 acquires the voice uttered by the user (YES in S21), the voice input unit 2 converts the acquired voice into voice data, outputs the voice data to the voice recognition unit 13, and the voice attached information of the voice data. Is output to the standby time prediction unit 16.

Subsequently, the standby time prediction unit 16 predicts the standby time (S22). Specifically, the standby time predicting section 16 receives the audio data, by using at least one of the data amount and the speech time of the speech data, calculates the predicted waiting time T _b. The standby time predicting unit 16 outputs the predicted standby time to the field connection operation control unit 17 (field connection operation determination unit 71) and the correction execution unit 22b. Subsequently, the field joining operation determination unit 71 performs a field joining operation determination process (S23). Note that the details of the joining operation determination process will be described later. The field connection operation determination unit 71 outputs, to the field connection operation execution unit 72, field connection operation information indicating the field connection operation determined to be executed by the voice interaction apparatus 10b. Then, the place joining operation execution unit 72 causes the voice interaction device 10b to execute the place joining operation according to the received place joining operation information (S24). When the execution of the joining operation is completed, the joining operation execution unit 72 notifies the modification execution unit 22b to that effect.

  On the other hand, the voice recognition unit 13 performs voice recognition processing (S25). Specifically, when receiving the voice data, the voice recognition unit 13 performs voice recognition processing on the voice data and outputs the voice recognition result to the response generation unit 14. Subsequently, the response generation unit 14 generates response information (S26). Specifically, the response generation unit 14 generates response information according to the received voice recognition result and outputs the response information to the correction execution unit 22b.

  As shown in FIG. 10, the processes in steps S22, S23, and S24 and the processes in steps S25 and S26 are performed in parallel. That is, when only one of the response information and the notification that the execution of the joining operation has been completed is received, the response correction unit 12b waits until the other is received. Then, when receiving the notification and the response information, the response correction unit 12b (correction necessity determination unit 21b) determines whether or not at least one of the utterance and the operation has been executed as the joining operation (S27). Specifically, the correction necessity determination unit 21b determines that the received connection operation information is No. in the connection operation information illustrated in FIG. It is determined whether the operation information is one field connection operation information or the other field connection operation information. Then, the determination result is output to the correction execution unit 22b.

The connection operation information received by the correction necessity determination unit 21b is No. If it is determined that the field joint operation information other than 1 (YES at S27), the correction execution unit 22b receives the determination result, the additional responses for predicted waiting time T _b received from the standby time predicting section 16 And the response information is corrected (S28). Specifically, correction execution unit 22b refers to the response modification table 61 stored in the storage unit 6, a specific additional response information associated with the waiting time matches the predicted waiting time T _b received To do. Then, the identified additional response information is read out and added to the response information received from the response generation unit 14, thereby correcting the response information. Then, the correction execution unit 22b outputs the corrected response information to the response execution unit 15.

  On the other hand, the connection operation information received by the correction necessity determination unit 21b is No. When it is determined that the information is the connection operation information of No. 1 (NO in S27), when the correction execution unit 22b receives the determination result, the response execution unit 15 does not correct the response information received from the response generation unit 14. Output to. That is, the process in step S28 described above is omitted.

  Finally, the response execution unit 15 causes the voice interaction device 10b to execute a response (S29). Specifically, the response execution unit 15 controls the voice output unit 4 to output voice according to the received response information, or controls the drive unit 5 to operate the movable part of the voice interaction device 10b. Or Thus, the response execution process ends.

  Next, the flow of the field connection operation determination process executed by the field connection operation determination unit 71 will be described with reference to FIG. FIG. 11 is a flowchart showing an example of the flow of the jointing operation determination process in the flowchart of FIG. In the flowchart of FIG. 11, it is assumed that the field connection operation information included in the field connection operation table 63 is associated with a field connection operation time that is generally equal to a standby time.

First, the field joint operation determining unit 71 receives the prediction waiting time T _b which waiting time prediction unit 16 predicts, reading a field joint operation table 63, the subtraction obtained by subtracting the field connecting operation time from the predicted waiting time T _b A value _Tc is calculated (S31). Subsequently, the field connection operation determination unit 71 refers to the field connection operation table 63 using the calculated subtraction value T _c and the first allowable time X, and determines whether there is field connection operation information satisfying 0 ≦ T _c ≦ X. It is determined whether or not (S32).

When there is field connection operation information satisfying 0 ≦ T _c ≦ X (YES in S32), the field connection operation determination unit 71 indicates the field connection operation indicated by one of the field connection operation information by the voice interactive device 10b. It is determined to perform the joining operation when performing (S33). Specifically, the field connection operation information having a smaller value of T _c is selected from the field connection operation information satisfying 0 ≦ T _c ≦ X. Then, the field connection operation determination unit 71 outputs the selected field connection operation information to the field connection operation execution unit 72.

On the other hand, when there is no field joining operation information that satisfies 0 ≦ T _c ≦ X (NO in S32), the field joining operation determining unit 71 calculates the sign change value −T _c from the subtraction value T _c, and the sign change value− By using _Tc and the second permissible time Y, it is determined whether or not there is connection operation information satisfying 0 ≦ −T _c ≦ Y by referring to the connection operation table 63 (S34).

If there is field connection operation information satisfying 0 ≦ −T _c ≦ Y (YES in S34), the field connection operation determination unit 71 displays the field connection operation indicated by one of the field connection operation information as the voice interactive device 10b. Is determined to be connected (S35). Specifically, among the field connection operation information satisfying 0 ≦ −T _c ≦ Y, the field connection operation information having a smaller value of −T _c is selected.

On the other hand, when there is no field connection operation information satisfying 0 ≦ −T _c ≦ Y (NO in S34), the field connection operation information indicating the field connection operation executed by the voice interactive device 10 by combining a plurality of field connection operations (S36). In this case, the spot joining operation execution unit 72 determines the order of operations of the received plurality of spot joining action information. This is the end of the joining operation determination process.

  In the present embodiment, the determination using both the first allowable time X and the second allowable time Y is performed. However, the determination using only one of the first allowable time X and the second allowable time Y is performed. You may go.

Further, in the present embodiment, 0 ≦ -T _c ≦ Y when no field joint operation information satisfying take, by combining a plurality of field joint operation information, was a place connecting operation information satisfying 0 ≦ -T _c ≦ Y However, it is good also as field joint operation information which satisfy | fills 0 <= _Tc <= X or 0 <=- _Tc <= Y by changing the field joint operation time matched with the field joint motion information.

Also, if the standby time predicting unit 16 takes time to generate the response information than the predicted waiting time T _b which is predicted waiting time prediction unit 16 uses the speech recognition result voice recognition unit 13 has performed again it may calculate the predicted waiting time T _b. Then, when the new predicted standby time becomes longer than the previous predicted standby time, the joining operation determining unit 71 may determine the joining operation again.

Further, in a case where the generation of early response information than the predicted waiting time T _b which waiting time prediction unit 16 predicts has been completed, if the field joint operation determiner 71 selects the plurality of field stitching operation, field The joining operation execution unit 72 may cancel the execution of the joining operation performed after that time. Also, if the generation of the early response information than the predicted waiting time T _b which waiting time prediction unit 16 predicts is completed, field joint operation execution unit 72 may accelerate the field joint operation time field joint action to perform.

Further, in the present embodiment, the field joining operation determination unit 71 selects the smaller subtraction value T _c (or sign change value −T _c ) when there are a plurality of field joining operation information that satisfy the conditions. However, it is not limited to this example. For example, if there is a column that stores history information indicating the date and time when each place-joining operation was last executed in the place-joining operation table 63, and there are multiple pieces of place-joint action information that satisfy the condition, the date and time when the history information is older May be selected.

  In addition, when the type of the field transition operation determined by the field transition operation determination unit 71 is “speech”, the field transition operation information in which the field transition operation time is shorter than the determined field transition operation and the type is “gesture” is displayed. It is also possible to select and combine the two pieces of joining operation information and output them to the joining operation execution unit 72. Similarly, when the determined type of the field transition operation is “gesture”, the field transition operation time is shorter than the determined field transition operation and the type is “speech”, and The joining operation information may be combined and output to the joining operation execution unit 72. For example, as shown in FIG. In the case where it is determined that the field connection operation (type: gesture, “get up” operation) indicated by the field connection operation information of 7 is the field connection operation to be executed by the voice interactive device 10b, the field connection operation determination unit 71, for example, No. 9 shown in FIG. 4 is further determined as a field connection operation executed by the voice interaction device 10b, and the field connection operation information is stored in the field connection operation (type: voice, “speak for a moment”). The data is output to the linking operation execution unit 72. Upon receiving this information, the spot-linking operation execution unit 72 causes the voice interaction apparatus 10b to execute “an operation to“ wake up ”while speaking” “Please wait a moment”. As a result, variations in the joining operation can be increased and the user can be prevented from getting bored.

Also, in the present embodiment, the predicted waiting time T _b which waiting time prediction unit 16 predicts, but the time and that is included in the response modification table 61 had identified the additional response information only on condition that matches, this example It is not limited to. For example, the correction execution unit 22b selects the additional response information in which the connection operation attribute indicating the category of the connection operation determined to be executed by the voice interaction apparatus matches the additional response attribute described in the second embodiment. , from the selected additional response information, time and included in the response modification table 61 and the predicted waiting time T _b may identify additional response information that matches.

In this case, the field connection operation information included in the field connection operation table 63 is associated with the field connection operation attribute indicating the category of the content of each field connection operation, and the response correction table 61 is stored in the storage unit 6b. Instead, the response correction table 61a described in the second embodiment is stored. The field connection operation determination unit 71 specifies the field connection operation attribute of the field connection operation that is determined to be executed by the voice interaction apparatus, and outputs the field connection operation attribute to the correction execution unit 22b. Correction execution unit 22b selects the received field joint operation attributes and additional response information associated with the additional response attribute matching, further from them time and included in the response modification table 61a and the predicted waiting time T _b The additional response information that matches is specified.

[Modification]
In Embodiment 1 described above, the voice recognition of the voice uttered by the user and the generation of response information for the voice are both performed by the voice dialogue control device (control unit 1). An external device (external server, not shown) that can communicate with the device 10 may perform the processing. That is, when the voice interaction apparatus 10 acquires voice, it converts it into voice data, and transmits the voice data to an external device via the communication unit 3. The external device performs voice recognition and response information generation, and transmits the response information to the voice interaction device 10. This modification can also be applied to the second and third embodiments.

  In the first embodiment described above, the control unit 1 may include a standby time prediction unit 16 instead of the standby time measurement unit 11, and may predict the standby time instead of measuring the standby time. This modification can also be applied to the second embodiment.

  Moreover, in Embodiment 1 mentioned above, although response information was corrected using the additional response information stored in the response correction table 61, correction of response information is not limited to this example. For example, at least a part of words included in the voice may be included in the correction response. Specifically, the correction execution unit 22 corrects the response information by using, as additional response information, a part of the voice data recognized by the voice recognition unit 13 that is cut out, edited, or summarized. Good. For example, when the voice uttered by the user, that is, the voice data recognized by the voice recognition unit 13 is “What is the weather today?”, The correction execution unit 22 receives the voice recognition result from the voice recognition unit 13, Cut out "Today's weather". Then, the correction execution unit 22 adds the voice data “Today's weather” to the voice data (response information) “Sunny” received from the response generation unit 14. Response information “It's sunny” is generated. This modification can also be applied to the second and third embodiments.

  Further, the correction condition is not limited to the example of each embodiment described above. For example, the response may be corrected when communication with an external device becomes impossible or when information acquired by a sensor (not shown) provided in the voice interaction device satisfies a predetermined condition. Further, the modification of the response is not limited to adding the additional response information to the response information. For example, the content of the response information may be changed. The content indicated by the additional response information may be executed before the response indicated by the response information, or may be executed after the response.

[Example of software implementation]
The voice interaction control devices of the voice interaction devices 10, 10a, and 10b, that is, the control units 1, 1a, and 1b may be realized by logic circuits (hardware) formed in an integrated circuit (IC chip) or the like. It may be realized by software using a CPU (Central Processing Unit).

  In the latter case, the voice interaction device 10 includes a CPU that executes instructions of a program that is software that realizes each function, and a ROM (Read Only Memory) in which the program and various data are recorded so as to be readable by a computer (or CPU). Alternatively, a storage device (these are referred to as “recording media”), a RAM (Random Access Memory) that expands the program, and the like are provided. And the objective of this invention is achieved when a computer (or CPU) reads the said program from the said recording medium and runs it. As the recording medium, a “non-temporary tangible medium” such as a tape, a disk, a card, a semiconductor memory, a programmable logic circuit, or the like can be used. The program may be supplied to the computer via an arbitrary transmission medium (such as a communication network or a broadcast wave) that can transmit the program. The present invention can also be realized in the form of a data signal embedded in a carrier wave in which the program is embodied by electronic transmission.

[Summary]
The voice dialogue control device (control unit 1) according to the first aspect of the present invention acquires the above-mentioned voice and the response generation unit (14) that generates a response to the voice uttered by the user, which is executed by the voice dialogue device (10). A determination unit (correction necessity determination unit 21) for determining whether or not a correction condition for determining whether or not the response needs to be corrected is satisfied during a waiting time until the response can be output; When the determination unit determines that the correction condition is satisfied, a correction unit (correction execution unit 22) that generates a correction response that corrects the response generated by the response generation unit, and the correction unit that generates the correction response. A response execution unit (15) for causing the voice interaction device to execute a correction response.

  According to the above configuration, since the response is corrected when the correction condition is satisfied from when the voice is acquired until the response can be output, the response can be corrected in a situation where the response should be corrected. . For example, when a time is required until a response can be output, this can be expressed by the response (additional audio that apologizes for the time required) can be output. Therefore, the flexibility of communication between the user and the voice interaction device can be improved.

  The voice interaction control device according to aspect 2 of the present invention further includes a standby time measuring unit (11) that measures the standby time in the above aspect 1, and the determination unit is the standby time measured by the standby time measuring unit. May exceed the predetermined time, it may be determined that the correction condition is satisfied.

  According to the above configuration, the standby time is measured, and when the measured standby time exceeds a predetermined time, it is determined that the correction condition is satisfied, so that it takes time until the response can be output. It can be determined accurately. Thereby, when it takes time until the response can be output, this can be expressed by the response.

  In the voice interaction control device according to aspect 3 of the present invention, in the aspect 2, the correction unit uses the correction content information indicating the correction content of the response associated with the time information corresponding to the standby time. The response may be modified.

  According to said structure, since a response is corrected using the correction content information matched with the time information corresponding to standby time, the response according to the length of standby time can be corrected. For example, when the standby time becomes long, an additional sound indicating that the user has been thinking for a long time is output. That is, since it is possible to respond according to the length of the waiting time, it is possible to prevent the user from feeling stress in communication with the voice interactive apparatus.

  The voice interaction control device according to aspect 4 of the present invention further includes a standby time prediction unit (16) that predicts the standby time in the above aspect 1, and the determination unit predicts standby that is predicted by the standby time prediction unit. When the time exceeds a predetermined time, it may be determined that the correction condition is satisfied.

  According to the above configuration, the standby time is predicted, and when the predicted standby time exceeds a predetermined time, it is determined that the correction condition is satisfied, so that it takes time until the response can be output. It can be determined accurately. Thereby, when it takes time until the response can be output, this can be expressed by the response.

  In the voice interaction control device according to aspect 5 of the present invention, in the aspect 4, the correction unit uses correction content information indicating the correction content of the response associated with time information corresponding to the predicted waiting time. Thus, the response may be corrected.

  According to said structure, since a response is corrected using the correction content information matched with the time information corresponding to prediction waiting time, the response according to the length of the estimated waiting time can be corrected. . For example, when the predicted standby time is long, an additional sound indicating that the user has been thinking for a long time is output. That is, since a response according to the predicted length of the standby time can be made, it is possible to prevent the user from feeling stress in communication with the voice interactive apparatus.

  The voice conversation control device according to aspect 6 of the present invention further includes a voice attribute specifying unit (speech recognition unit 13) that specifies a voice attribute indicating a category of the content of the voice in the mode 3 or 5, and the correction content. The information is further associated with a response attribute indicating the category of the correction content, and the correction unit is configured to associate the correction attribute associated with the response attribute corresponding to the voice attribute specified by the voice attribute identification unit. The response may be modified using content information.

  According to the above configuration, the response is corrected using the correction content information associated with the time information corresponding to the standby time or the predicted standby time and associated with the response attribute corresponding to the voice attribute. The device can make a response with a more appropriate modification to the voice uttered by the user.

  The voice interaction control device according to aspect 7 of the present invention further includes a field connection operation determination unit (71) that determines a field connection operation to be executed by the voice interaction device during the standby time in the aspect 1, and the determination unit May determine that the correction condition is satisfied when the place-joining movement determination unit determines that the voice interactive apparatus performs at least one of voice output and gesture as the spot-joining action.

  Since the connection operation is an operation that fills the waiting time from when the voice is acquired until the response can be output, the voice interaction device executes at least one of the voice output and the operation as the operation. It takes time to generate a response. Here, according to the above configuration, when the voice interaction device performs at least one of voice output and gesture as the connecting operation, it is determined that the correction condition is satisfied, and thus it takes time to generate a response. Can be accurately determined.

  The voice interaction control device according to aspect 8 of the present invention further includes a standby time prediction unit that predicts the standby time in aspect 7, and the correction unit corresponds to the predicted standby time predicted by the standby time prediction unit. The response may be corrected using correction content information indicating the correction content of the response, which is associated with the time information.

  According to said structure, since a response is corrected using the correction content information matched with the time information corresponding to prediction waiting time, the response according to the length of the estimated waiting time can be corrected. . That is, since a response according to the predicted length of the standby time can be made, it is possible to prevent the user from feeling stress in communication with the voice interactive apparatus.

  In the voice interaction control device according to aspect 9 of the present invention, in the aspect 8, the correction content information is further associated with a response attribute indicating the category of the correction content, The field connection operation attribute indicating the determined field connection operation category is specified, and the correction unit is associated with the response attribute corresponding to the field connection operation attribute specified by the field connection operation determination unit. The response may be modified using content information.

  According to the above configuration, since the response is corrected using the correction content information that is associated with the time information corresponding to the standby time and is associated with the response attribute corresponding to the jumping operation attribute, the voice interactive apparatus is , It is possible to make a response with corrections suitable for the connecting operation performed.

  In the voice interaction control device according to aspect 10 of the present invention, in any one of the aspects 1 to 9, the correction unit may include at least a part of words included in the voice in the correction response.

  When it takes time until a response is generated after acquiring the voice uttered by the user, it is desirable to include the content of the voice in the response. So, according to said structure, at least one part of the word contained in an audio | voice is included in a correction response. As a result, communication between the user and the voice interactive apparatus can be smoothly advanced. Note that including at least a part of the words included in the voice in the correction response means, for example, “sunny weather”, which is a response to the voice “What is the weather today?” “Today's weather is sunny” is to correct. Further, the response may be corrected by using a content obtained by editing the voice, that is, a part of a word included in the voice and reconstructed. For example, the words “Today” and “Ha” are extracted from the voice data “What is the weather today?”, The voice data “Today is” is reconstructed, and the response “It ’s sunny” It's fine. "

  The control method of the voice interaction control device according to aspect 11 of the present invention includes a response generation step for generating a response to the voice uttered by the user, which is executed by the voice interaction device, and the response can be output after acquiring the voice. A determination step for determining whether or not a correction condition for determining whether or not the response needs to be corrected is satisfied during the waiting time until the time is determined, and when it is determined in the determination step that the correction condition is satisfied A correction step for generating a correction response by correcting the response generated in the response generation step, and a response execution step for causing the voice interaction apparatus to execute the correction response generated in the correction step. According to this control method, the same function and effect as those of the spoken dialogue control apparatus according to aspect 1 are obtained.

  The voice interaction apparatus according to aspect 12 of the present invention may include the voice interaction control apparatus according to any one of aspects 1 to 10 described above. According to the above configuration, this voice interaction apparatus can improve the flexibility of communication with the user.

  The voice conversation control device according to each aspect of the present invention may be realized by a computer. In this case, the voice conversation control device is operated by causing the computer to operate as each unit (software element) included in the voice dialogue control device. The control program of the voice interaction control device that realizes the above in a computer and a computer-readable recording medium on which the control program is recorded also fall within the scope of the present invention.

  The present invention is not limited to the above-described embodiments, and various modifications are possible within the scope shown in the claims, and embodiments obtained by appropriately combining technical means disclosed in different embodiments. Is also included in the technical scope of the present invention. Furthermore, a new technical feature can be formed by combining the technical means disclosed in each embodiment.

  INDUSTRIAL APPLICABILITY The present invention can be used for a voice dialogue control device for controlling a voice dialogue device that responds to a user's utterance.

  1, 1a, 1b control unit (voice dialogue control device), 10, 10a, 10b voice dialogue device, 11 standby time measurement unit, 13 voice recognition unit (voice attribute identification unit), 14 response generation unit, 15 response execution unit, 16 standby time prediction unit, 21, 21b correction necessity determination unit (determination unit), 22, 22a, 22b correction execution unit (correction unit), 71 spot connection operation determination unit, S4 response generation step, S6 determination step, S7 correction Step, S8 Response execution step

Claims (4)

A response generation unit that generates a response to the voice uttered by the user, executed by the voice interaction device;
A determination unit that determines whether or not a correction condition for determining whether or not the response needs to be corrected is satisfied in a waiting time from when the voice is acquired until the response can be output;
When the determination unit determines that the correction condition is satisfied, a correction unit that generates a correction response that corrects the response generated by the response generation unit;
A response execution unit that causes the voice interaction device to execute the correction response generated by the correction unit;
A standby time prediction unit for predicting the standby time ,
The determination unit, when the predicted waiting time the standby time predicting section predicts exceeds a predetermined time, speech dialog control device which is characterized that you determined the adjustment conditions are satisfied. The voice according to claim 1 , wherein the correction unit corrects the response using correction content information indicating correction content of the response associated with time information corresponding to the standby time. Dialog control device. A response generation step for generating a response to the voice uttered by the user, executed by the voice interaction device;
A determination step of determining whether or not a correction condition for determining whether or not the response needs to be corrected is satisfied in a waiting time from when the voice is acquired until the response can be output;
A correction step for generating a correction response in which the response generated in the response generation step is corrected when it is determined in the determination step that the correction condition is satisfied;
A response execution step for causing the voice interaction apparatus to execute the correction response generated in the correction step;
And the step of predicting the above-mentioned waiting time, only including,
In the determination step, the control method of the spoken dialogue control apparatus, wherein the correction condition is determined to be satisfied when the predicted standby time predicted in the step of predicting the standby time exceeds a predetermined time . Voice dialogue system, characterized in that it comprises a speech dialog control device according to claim 1 or 2.

Images