JP3576511B2 - Voice interaction device - Google Patents

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a voice interaction device using a voice recognition technology and a voice synthesis technology.
[0002]
[Prior art]
2. Description of the Related Art In recent years, various devices have been equipped with a voice interaction device that achieves a desired purpose by performing voice recognition and dialogue, such as a navigation device that assists an operation without manual input. It is installed in.
[0003]
As this type of voice interactive device, for example, when mounted on a navigation device, as shown in FIG. 23, it is necessary to continue a dialog with a user to display and set a destination. Can be done.
[0004]
[Problems to be solved by the invention]
However, in such a conventional voice interactive device, for example, as shown in FIG. 23, a search is made for a XX golf course in Chiba using an item search function for setting a destination of a navigation device. In such a case, if the user does not know that the XX golf course is located in Chiba Prefecture, the conversation cannot be continued, and the search becomes impossible.
[0005]
In other words, if the user does not have an answer to the question asked by the dialogue device, the dialogue will be interrupted. There was a problem that the user's purpose could not be achieved because the dictionary could not be selected.
[0006]
SUMMARY OF THE INVENTION The present invention has been made to solve such a problem, and a voice dialogue apparatus capable of continuing a dialogue and achieving an object even when a user cannot correctly answer a question asked by the apparatus side. To provide.
[0007]
[Means for Solving the Problems]
The voice interaction device of the present invention In a voice interaction device that performs a dialogue in response to a voice uttered by a user, input Said that Voice recognition means for recognizing voice; Said Dictionary storage means for storing a speech recognition dictionary for each level of dialogue; Said Dictionary preparation means for preparing a dictionary according to the dialogue; For the user Prompt vocalization response Response voice output means for outputting voice; A next response voice is created using the input voice, and the next response voice is generated using the past response voice or the previously input voice when the content indicates that the voice is unknown. To create The voice recognition dictionary is prepared by the dictionary preparation means, and the response voice output means The response And a dialogue control unit for outputting a voice.
[0008]
With such a configuration, when an instruction is made in accordance with the user's speech (voice) recognized by the speech, a speech recognition dictionary is prepared from the dictionary storage means, and a voice prompting the user to speak, In other words, while the dialogue is continued by outputting a question voice or a response voice, etc., if the voice content of the voice-recognized user cannot determine the next instruction, a speech recognition dictionary effective for continuing the dialogue. Is prepared from the dictionary storage means, and then the dialogue is continued by outputting a question voice and a response voice. Therefore, even when the user cannot respond or can only give an ambiguous answer, the dialog can be continued without interruption and the object can be achieved.
[0009]
The dictionary preparing means of the voice dialogue apparatus of the present invention prepares a dictionary necessary for dialogue by selectively combining the voice recognition dictionaries in the dictionary storage means, and the dialogue control means is recognized by the voice recognition means. Was Of the content indicating that the voice is unknown In some cases, the speech recognition dictionary in the dictionary storage unit of the hierarchy to which the next transition is likely to be made is selectively combined and prepared with the dictionary preparation unit.
[0010]
With such a configuration, when the next instruction cannot be determined in a meaning that cannot be answered because the content of the voice of the user whose voice has been recognized is unknown, all of the possibilities that may be used when the dialogue is continued. The voice recognition dictionary is prepared by being selected from the dictionary storage means and combined, and then the dialogue is continued by outputting a question voice, a response voice, and the like. Therefore, even when the user cannot respond because he does not know the answer, the dialog can be continued without interruption, and the object can be achieved.
[0011]
The dictionary preparing means of the voice dialogue apparatus of the present invention selects the voice recognition dictionary in the dictionary storage means to prepare a dictionary necessary for a dialogue, and the dialogue control means is recognized by the voice recognition means. Of the content indicating that the voice is unknown In some cases, a voice for asking another question is output to the response voice output unit, and the dictionary preparation unit is configured to select and prepare the voice recognition dictionary required for a conversation from the dictionary storage unit.
[0012]
According to such a configuration, when the next instruction cannot be determined because the content of the voice of the user whose voice has been recognized cannot be answered because it is unknown, another question is asked, and the voice recognition dictionary corresponding to the answer is issued. By being prepared by being selected from the dictionary storage means, a question voice, a response voice, and the like are output, and the dialogue is continued. Therefore, even when the user cannot respond because he does not know the answer, the dialog can be continued without interruption, and the object can be achieved.
[0013]
The dictionary preparing means of the voice dialogue apparatus of the present invention prepares a dictionary necessary for dialogue by selectively combining the voice recognition dictionaries in the dictionary storage means, and the dialogue control means is recognized by the voice recognition means. When the voice that has been heard has a content meaning unknown for the first time, another voice is output to the response voice output means. Answer voice To cause the dictionary preparation means to select and prepare the speech recognition dictionary required for a conversation from the dictionary storage means, and to be recognized by the speech recognition means. When the user's voice is unknown Is continued, the speech recognition dictionaries in the dictionary storage means of all the hierarchies which may transition next are selectively combined with the dictionary preparation means to prepare them.
[0014]
With such a configuration, when the voice of the user who cannot determine the next instruction for the first time is recognized because the content is unknown and cannot be answered, another question is asked and a voice recognition dictionary corresponding to the answer is given. Is prepared by being selected from the dictionary storage means, and the dialogue is continued by outputting question voices and response voices. However, since the content of the user's voice is unknown even in the next voice recognition, If the instruction cannot be confirmed, all the speech recognition dictionaries that may be used when the dialogue is continued are prepared by being selected from the dictionary storage means and combined, and then the question speech and response The dialogue is continued by outputting a voice or the like. Therefore, even when the user cannot respond because he does not know the answer, the dialog can be continued without interruption, and the object can be achieved.
[0015]
The dictionary preparing means of the voice dialogue apparatus of the present invention prepares a dictionary necessary for dialogue by selectively combining the voice recognition dictionaries in the dictionary storage means, and the dialogue control means is recognized by the voice recognition means. When the content indicates that the sound is ambiguous, The voice recognition dictionary is provided to the dictionary preparing means so as to create a next response voice using the past response voice or the voice input in the past. It has a configuration to prepare.
[0016]
With such a configuration, when the content of the voice of the user whose speech has been recognized cannot be determined with the meaning of the ambiguous answer, the corresponding speech recognition dictionary is removed when the words representing the ambiguity are removed. In addition, a speech recognition dictionary that is conceptually similar is prepared by being selected from the dictionary storage means and combined, and then the dialogue is continued by outputting a question voice, a response voice, and the like. Therefore, even when the user does not know the correct answer, the dialog can be continued without interruption, and the object can be achieved.
[0017]
The dictionary preparing means of the voice dialogue apparatus of the present invention prepares a dictionary necessary for dialogue by selectively combining the voice recognition dictionaries in the dictionary storage means, and the dialogue control means is recognized by the voice recognition means. When the voice includes a plurality of contents, a plurality of the speech recognition dictionaries in the dictionary storage means corresponding to the plurality of contents are selectively coupled to the dictionary preparation means to prepare.
[0018]
With such a configuration, when the next instruction cannot be determined with the meaning of the user whose voice has been recognized and includes a plurality of contents, the voice recognition dictionary corresponding to each content is selected from the dictionary storage means and combined. Then, the dialogue is continued by outputting a question voice, a response voice, and the like. Therefore, even when the user does not know the correct answer, the dialog can be continued without interruption, and the object can be achieved.
[0019]
The dictionary preparing means of the voice dialogue apparatus of the present invention selects one of the voice recognition dictionaries in the dictionary storage means to prepare a dictionary necessary for a dialogue, and the dialogue control means recognizes the speech by the voice recognition means. When the input speech includes a plurality of contents, the dictionary preparation unit selects and prepares the speech recognition dictionary in the dictionary storage unit corresponding to one content included in the contents, and asks the response speech output unit to answer the question. By confirming the correctness of the dialogue with the voice recognized by the voice recognition unit, if the voice is recognized by the voice recognition unit, and if it is wrong, in the dictionary storage unit corresponding to other contents included in the voice The voice recognition dictionary is selected and prepared by the dictionary preparation means.
[0020]
With such a configuration, when the voice of the user whose voice has been recognized cannot be determined with the semantic content including a plurality of contents, a voice recognition dictionary corresponding to one content is first selected and prepared from the dictionary storage means. Then, when another question is asked, it is confirmed whether or not one of the contents is satisfactory. If it is incorrect, a speech recognition dictionary corresponding to the other content is selected and prepared from the dictionary storage means. Then, the dialogue is continued by outputting a question voice, a response voice, and the like. Therefore, even when the user does not know the correct answer, the dialog can be continued without interruption, and the object can be achieved.
[0021]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, the present invention will be described with reference to the drawings. FIG. 1 to FIG. 6 are views showing a first embodiment of the voice interaction apparatus of the present invention.
[0022]
First, the device configuration will be described. In FIG. 1, a voice interaction device 10 includes a voice recognition unit 11 for recognizing a voice input by a user, a dialog control unit 12 for controlling a voice interaction with the user, and a hierarchy (type and progress) of the dialog. And the like, and a dictionary storage unit 13 in which necessary speech recognition dictionaries are stored for all dialog layers, and one or more speech recognition dictionaries stored in the dictionary storage unit 13 according to a command from the dialog control unit 12. A dictionary selection / coupling unit (dictionary preparation means) 14 for creating a voice recognition dictionary used by the voice recognition unit 11 by combining the data, and a question voice or a response voice prompting the user to generate a voice according to a command from the dialog control unit 12. , A response voice storage unit 16 for storing a plurality of voices used in the response voice output unit 15, and an unknown table in which words meaning unknown are registered as items. A word dictionary 17, and an unknown expression word determining unit 18 that determines whether or not the speech recognition result is unknown by referring to the unknown expression word dictionary 17 in response to an inquiry from the dialog control unit 12. It is mounted on a navigation device to assist operations such as search and destination setting by voice input.
[0023]
Next, the processing operation of the speech dialogue apparatus of the present invention will be described with reference to the flowchart of the dialogue shown in FIG.
[0024]
First, when a voice dialogue is started by an instruction of a user (a user of the navigation device), the dialogue control unit 12 instructs the dictionary selection / connection unit 14 to create a dictionary including words representing a genre of search. In response to this command, the dictionary selecting / combining unit 14 creates a speech recognition dictionary including words representing a search genre from the speech recognition dictionary storage unit 13 as shown in FIG.
[0025]
Next, the dialogue control unit 12 instructs the response voice output unit 15 to output a message prompting the user to speak words. In response to this command, the response voice output unit 15 selects a message "What is your need?" From the response voice storage unit 16 and presents it to the user.
[0026]
Next, the dialog control unit 12 instructs the speech recognition unit 11 to execute speech recognition using the dictionary created by the dictionary selection and combination unit 14. When the user who has heard the message "What is your need?" Utters "Facility search." To search for a facility and inputs it to the voice interactive device 10, the input voice is Recognized by the voice recognition unit 11, “Facility search.” Is selected as a command as a recognition result, and output to the dialog control unit 12. Based on this result, the dialogue control unit 12 also sends words indicating the genre of search to the dictionary selection / connection unit 14 together with words that may be uttered when the user does not know the type of facility, such as “I do not understand.” Instructs the dictionary to be included. In response to this command, the dictionary selection / coupling unit 14 selects (creates) a speech recognition dictionary including words indicating the search genre and words such as “I do not understand.” From the speech recognition dictionary storage unit 13 as shown in FIG. Do.
[0027]
Next, the dialog control unit 12 instructs the response voice output unit 15 to output a message prompting the user to utter the type of facility. In response to this command, the response voice output unit 15 selects a message “Please tell us the type of facility” from the response voice storage unit 16 and presents it to the user.
[0028]
Next, the dialog control unit 12 instructs the speech recognition unit 11 to execute speech recognition using the dictionary created by the dictionary selection and combination unit 14. When the user who hears the message "Please tell us the type of facility." Utters "Golf course." Is recognized by the voice recognition unit 11, and “Golf course.” Is selected as a search genre as a recognition result.
[0029]
Next, the dialogue control unit 12 may narrow down the location of the golf course and, if the user does not know the prefecture name where the golf course is located, such as “I do not know.” Instructs the creation of a dictionary composed of certain words. In response to this command, the dictionary selecting / combining unit 14 creates a speech recognition dictionary composed of a prefecture name and words such as "I do not understand" as shown in FIG. 5 from the speech recognition dictionary storage unit 13.
[0030]
Next, the dialogue control unit 12 instructs the response voice output unit 15 to output a message prompting the user to utter the name of the prefecture where the golf course is located. In response to this command, the response voice output unit 15 selects a message "Please talk about the name of the prefecture where the golf course is located" from the response voice storage unit 16 and presents it to the user. When the user who hears this message does not know the name of the prefecture where the golf course is located, he utters “I do not know.” And inputs it to the voice interaction device 10, and the input voice is recognized by the voice recognition unit 11. Then, "I do not understand." Is selected as the recognition result.
[0031]
Next, the dialog control unit 12 outputs this result to the unknown expression word determination unit 18. The unknown expression word determination unit 18 receives this result, refers to the unknown expression word dictionary 17 having words indicating unknown as shown in FIG. 6 as an item, determines whether the word is unknown, and determines whether the word is unknown. The result is output to the dialog control unit 12. In this case, since "I do not understand." Is determined as a word representing unknown, in response to this determination result, the dialogue control unit 12 causes the dictionary selection connection unit 14 to input all the dictionaries of the golf course divided by prefecture name. Order creation of a combined dictionary. In response to this command, the dictionary selection / coupling unit 14 extracts all the golf course dictionaries classified by prefecture name from the voice recognition dictionary storage unit 13 and creates a voice recognition dictionary by combining them.
[0032]
Next, the dialog control unit 12 instructs the response voice output unit 15 to output a message prompting the user to utter the name of the golf course. In response to this command, the response voice output unit 15 selects a message “Please tell us the name of the golf course” from the response voice storage unit 16 and presents it to the user. When the user who hears this message utters the name of the golf course "XX golf course." And inputs it to the voice interaction device 10, the input voice is recognized by the voice recognition unit 11, and as a recognition result. , "XX golf course." Is selected, and the search target is determined.
[0033]
Next, the dialogue control unit 12 instructs the response voice output unit 15 to present the determined search target “XX golf course.” To the user. In response to this command, the response voice output unit 15 combines the content stored in the response voice storage unit 16 with “XX golf course.” And displays a message “Display a map of golf course.” Create and present to the user.
[0034]
With the above operation, a map such as a destination to be searched can be displayed on the display screen of the navigation device.
[0035]
As described above, in the first embodiment, by providing the dictionary selecting / combining unit 14 and the unknown expression determination unit 18, the user can say, for example, “I do not know” without knowing the name of the prefecture where the golf course is located. Even if the next instruction cannot be determined because you answered the question during the dialogue, take out all the golf course dictionaries divided by prefecture name that would be used when continuing the dialogue By creating a combined voice recognition dictionary and performing voice recognition, a search target can be determined without interrupting the flow of voice dialogue. Therefore, even in the case of a response in which the user does not know the answer, the dialog can be continued without interruption, and the object can be achieved.
[0036]
Next, FIGS. 7 to 9 are diagrams showing a second embodiment of the voice interaction apparatus of the present invention. Note that the second embodiment is configured substantially in the same manner as the first embodiment described above. explain.
[0037]
First, the device configuration will be described. In FIG. 7, the voice interaction device 20 includes a voice recognition unit 11, a dialogue control unit 12, a dictionary storage unit 13, a response voice output unit 15, a response voice storage unit 16, an unknown expression word dictionary 17, and an unknown expression word dictionary. A dictionary selection unit (dictionary preparation means) 24 is provided in place of the dictionary selection and combination unit 14 in the first embodiment described above. By selecting one of the speech recognition dictionaries stored in the dictionary storage unit 13 according to a command from the control unit 12, a speech recognition dictionary used by the speech recognition unit 11 is created.
[0038]
Next, the processing operation of the speech dialogue apparatus of the present invention will be described with reference to the flowchart of the dialogue shown in FIG.
[0039]
First, in the same manner as in the first embodiment described above, a voice dialogue is started by a user's instruction, and the user responds to the message "What is your use?" When "Facility search." Is uttered and this "Facility search." Is selected as a command, a voice recognition dictionary containing words indicating a search genre and words such as "I do not understand." Once created, a message "Please tell us the type of facility." Is presented to the user.
[0040]
Then, when the user who hears the message "Please tell us the type of facility" does not understand the word indicating the genre to be searched, he utters "I do not understand." Thus, the input voice is recognized by the voice recognition unit 11, and "I do not understand" is selected as the recognition result.
[0041]
Next, the dialog control unit 12 outputs this result to the unknown expression word determination unit 18. The unknown expression word determining unit 18 receives this result, refers to the unknown expression word dictionary 17 having words indicating unknown as shown in FIG. 6 as items, and determines whether or not the word represents unknown. The result is output to the dialog control unit 12. In this case, "I don't understand." Is determined as a word indicating unknown, and the dialog control unit 12 responds to this determination result by using the dictionary selection unit 24 to narrow down the location of the facility. Command selection. In response to this command, the dictionary selection unit 24 selects a speech recognition dictionary composed of prefecture names as shown in FIG. 9 from the speech recognition dictionary storage unit 13.
[0042]
Next, the dialog control unit 12 instructs the response voice output unit 15 to output a message prompting the user to utter the name of the prefecture where the facility is located. In response to this command, the response voice output unit 15 selects a message "Please tell us the name of the prefecture where the facility is located" from the response voice storage unit 16 and presents it to the user. The user who hears this message utters the name of the prefecture where the facility is located, "Chiba prefecture", and inputs it to the voice interaction device 20. The input voice is recognized by the voice recognition unit 11, and "Chiba Prefecture" is selected. In response to this result, the dialog control unit 12 instructs the dictionary selection unit 24 to select a dictionary composed of facilities of all genres in Chiba Prefecture. In response to this command, the dictionary selection unit 24 selects a dictionary composed of facilities of all genres in Chiba from the speech recognition dictionary storage unit 13.
[0043]
Next, the dialogue control unit 12 instructs the response voice output unit 15 to output a message prompting the user to speak the name of the facility. In response to this command, the response voice output unit 15 selects a message "Please tell us the name of the facility in Chiba Prefecture" from the response voice storage unit 16 and presents it to the user. When the user who hears this message utters the name of the facility “ゴ ル フ golf course.” And inputs it to the voice interaction device 20, the input voice is recognized by the voice recognition unit 11, and as a recognition result, “XX golf course.” Is selected, and the search target is determined.
[0044]
Next, the dialogue control unit 12 instructs the response voice output unit 15 to present the determined search target “XX golf course.” To the user. In response to this command, the response voice output unit 15 combines the content stored in the response voice storage unit 16 with “XX golf course.” And displays a message “Display a map of golf course.” Create and present to the user.
[0045]
With the above operation, a map such as a destination to be searched can be displayed on the display screen of the navigation device.
[0046]
As described above, in the second embodiment, by providing the dictionary selecting unit 24 and the unknown expression determining unit 18, the user can be in a dialogue, for example, with “unknown” without knowing the name of the facility genre. Even if you can not determine the next instruction because you answered the question, if you know the prefecture name, by performing speech recognition using a dictionary containing facilities of all genres divided by prefecture name, The search target can be determined without interrupting the flow of the voice dialogue. Therefore, even in the case of a response in which the user does not know the answer, the dialog can be continued without interruption, and the object can be achieved.
[0047]
As another mode of the second embodiment, as shown in FIG. 10, in response to a message "Please tell us the type of facility." Is repeatedly input in response to the message "", a dictionary composed of facilities of all genres is selected in the dictionary selection unit 24, and "the name of the facility is entered." The user who hears this message utters the name of the facility "XX golf course." Is determined, and a message "Display a map of the XX golf course." Is presented to the user.
[0048]
Next, FIG. 11 to FIG. 14 are views showing a third embodiment of the voice interaction apparatus of the present invention. Since the third embodiment is configured substantially in the same manner as the above-described first embodiment, the same reference numerals are given to the same configurations using the drawings, and the characteristic portions will be described. explain.
[0049]
First, the device configuration will be described. In FIG. 11, the voice interaction device 30 includes a voice recognition unit 11, a dialogue control unit 12, a dictionary storage unit 13, a dictionary selection connection unit 14, a response voice output unit 15, and a response voice storage unit 16. In addition, an ambiguity expression word dictionary 37 and an ambiguity expression word determination unit 38 are provided in place of the unknown expression word dictionary 17 and the unknown expression word determination unit 18 in the above-described first embodiment. A part 39 is provided.
[0050]
In the vague expression word dictionary 37, words meaning vague are registered as items, and the vague expression word determination unit 38 refers to the vague expression word dictionary 37 in response to an inquiry of the dialog control unit 12, and performs voice recognition. It is determined whether the result expresses an ambiguous thing.
[0051]
The distance calculation unit 39 calculates whether or not the distance is included in the specific distance and selects a range to be searched. For example, for the designation of Tokyo, the neighboring prefectures Chiba and Saitama Prefecture, Kanagawa prefecture, Yamanashi prefecture.
[0052]
Next, the processing operation of the speech dialogue apparatus of the present invention will be described with reference to the flowchart of the dialogue shown in FIG.
[0053]
First, in the same manner as in the first embodiment described above, a voice dialogue is started by a user's instruction, and the user responds to the message "What is your use?" While saying "Facility search.", A speech recognition dictionary of a search genre as shown in FIG. 4 is created, and a message "Please tell us the type of facility" is presented to the user. When the user utters “Golf course.” And inputs it to the voice interaction device 30, the input voice is recognized by the voice recognition unit 11, and “Golf course” is selected as a search genre.
[0054]
Then, in order to narrow down the location of the golf course, the dialogue control unit 12 displays the prefecture name and the prefecture name where the user is located, such as "Kanaa." The dictionary selection / combination unit 14 is instructed to create a dictionary composed of words that may be uttered when the user only knows vaguely. In response to this command, the dictionary selection / coupling unit 14 uses the prefecture name as the main dictionary as shown in FIG. 13, and uses words such as "Kanaa." A recognition dictionary is selected from the speech recognition dictionary storage unit 13 and created.
[0055]
Next, the dialogue control unit 12 instructs the response voice output unit 15 to output a message prompting the user to utter the name of the prefecture where the golf course is located. In response to this command, the response voice output unit 15 selects a message "Please talk about the name of the prefecture where the golf course is located" from the response voice storage unit 16 and presents it to the user. The user who heard this message, In the Nth conversation hierarchy (where N is a natural number; the same applies hereinafter): Since the name of the prefecture where the golf course is located is ambiguous, by saying "Tokyo Kana." And inputting it to the voice interaction device 30, the input voice is recognized by the voice recognition unit 11, and as a recognition result, "Tokyo" I wonder if the city is. "
[0056]
Next, the dialog control unit 12 outputs this result to the ambiguous expression word determination unit 38. The ambiguous expression word judging unit 38 receives this result, and refers to the ambiguous expression word dictionary 37 having words indicating that it is ambiguous as shown in FIG. 14 as an item, and judges whether or not the word includes an ambiguous word. And outputs the result to the dialog control unit 12. In this case, “Kanaa.” Is determined as a word that represents ambiguity. In response to this determination result, the dialog control unit 12 calculates the distance of each prefecture to “Tokyo.” From which the words indicating ambiguity have been removed from the recognition result to the distance calculation unit 39, and determines the prefecture whose distance is close to Tokyo. Command to make a selection. In response to this command, the distance calculator 39 selects Chiba, Saitama, Kanagawa, and Yamanashi prefectures, and instructs the dictionary selection coupling unit 14 to combine the golf course dictionaries in which Tokyo is added to these four prefectures. According to this command, the dictionary selection / coupling unit 14 takes out and combines the dictionaries of the golf courses in Chiba, Saitama, Kanagawa, Yamanashi and Tokyo from the voice recognition dictionary storage unit 13, For the N + 1th conversation hierarchy Create a speech recognition dictionary.
[0057]
Next, the dialog control unit 12 instructs the response voice output unit 15 to output a message prompting the user to utter the name of the golf course. In response to this command, the response voice output unit 15 selects a message “Please tell us the name of the golf course” from the response voice storage unit 16 and presents it to the user. When the user who hears this message utters the name of the golf course "XX golf course." And inputs it to the voice interaction device 30, the input voice is recognized by the voice recognition unit 11, and as a recognition result. , "XX golf course." Is selected, and the search target is determined.
[0058]
Next, the dialogue control unit 12 instructs the response voice output unit 15 to present the determined search target “XX golf course.” To the user. In response to this command, the response voice output unit 15 combines the content stored in the response voice storage unit 16 with “XX golf course.” And displays a message “Display a map of golf course.” Create and present to the user.
[0059]
With the above operation, a map such as a destination to be searched can be displayed on the display screen of the navigation device.
[0060]
As described above, in the third embodiment, by providing the dictionary selecting / combining unit 14 and the ambiguous expression word judging unit 38, the user can know, for example, only vaguely the name of the prefecture where the golf course is located. Even if you couldn't determine the next instruction because you answered the question during the conversation, create a voice recognition dictionary that combines dictionaries of golf courses in prefectures that are close to Tokyo. By performing voice recognition, a search target can be determined without interrupting the flow of voice dialogue. Therefore, even if the user gets an ambiguous response without knowing the correct answer, the dialog can be continued without interruption and the object can be achieved.
[0061]
Next, FIG. 15 to FIG. 18 are views showing a fourth embodiment of the voice interaction apparatus of the present invention. Since the fourth embodiment is configured substantially in the same manner as the above-described third embodiment, the same reference numerals are given to the same configurations using the drawings, and the characteristic portions will be described. explain.
[0062]
First, the device configuration will be described. In FIG. 15, the voice interaction device 40 includes a voice recognition unit 11, a dialogue control unit 12, a dictionary storage unit 13, a dictionary selection connection unit 14, a response voice output unit 15, a response voice storage unit 16, A concept dictionary table 47 and a similar concept selection unit 48 are provided in addition to the expression word dictionary 37 and the ambiguous expression word determination unit 38, in addition to the configuration of the third embodiment.
[0063]
The concept dictionary table 47 is set in advance by associating similar concepts as shown in FIG. 17, and the similar concept selecting unit 48 determines which one to adopt with reference to the concept dictionary table 47. It is supposed to.
[0064]
Next, the processing operation of the speech dialogue apparatus of the present invention will be described with reference to the flowchart of the dialogue shown in FIG.
[0065]
First, in the same manner as in the third embodiment described above, a voice dialogue is started by a user's instruction, and the user responds to the message "What is your need?""Facilitysearch." Is spoken, and this "facility search." Is selected as a command.
[0066]
Then, when this "facility search." Is instructed, the dialog control unit 12 narrows down the type of the facility, and the name of the facility and "Kanaa." When the user only knows the type of the facility vaguely, he instructs the dictionary selection / combination unit 14 to create a dictionary composed of words likely to be uttered. In response to this command, the dictionary selecting / combining unit 14 uses the name of the facility as the main dictionary as shown in FIG. 18 and uses words such as "Kanaa." A recognition dictionary is selected from the speech recognition dictionary storage unit 13 and created.
[0067]
Next, the dialogue control unit 12 instructs the speech recognition unit 11 to execute speech recognition using the dictionary created by the dictionary selection and combination unit 14, and instructs the response speech output unit 15 to the user. Instructs to output a message prompting the utterance of the facility type. In response to this command, the response voice output unit 15 reads from the response voice storage unit 16 "Please tell us the type of facility." For the Nth conversation hierarchy Select a message and present it to the user. The user who heard the message "Please tell us the type of facility." , In the Nth conversation hierarchy, By uttering "Kana zoo" as a word indicating the genre to be searched and inputting it to the voice interaction device 40, the input voice is recognized by the voice recognition unit 11, and as a recognition result, "Kana zoo. Is selected.
[0068]
Next, the dialog control unit 12 outputs this result to the ambiguous expression word determination unit 38. The ambiguous expression word judging unit 38 receives this result, and refers to the ambiguous expression word dictionary 37 having words indicating that it is ambiguous as shown in FIG. 14 as an item, and judges whether or not the word includes an ambiguous word. And outputs the result to the dialog control unit 12. In this case, “Kanaa.” Is determined as a word that represents ambiguity. In response to this determination result, the dialogue control unit 12 sends a similar concept selection table 48 to the concept dictionary table 47 as shown in FIG. , And the result is output to the dialogue control unit 12. The dialog control unit 12 stores this result together with the recognition result "zoo."
[0069]
Next, the dialog control unit 12 instructs the response voice output unit 15 to output a message prompting the user to utter the name of the prefecture where the facility is located. In response to this command, the response voice output unit 15 selects a message "Please tell us the name of the prefecture where the facility is located" from the response voice storage unit 16 and presents it to the user. The user who hears this message utters "Osaka prefecture." And inputs it to the voice interaction device 40. The input voice is recognized by the voice recognition unit 11, and as a recognition result, "Osaka prefecture." Is selected.
[0070]
Next, the dialogue control unit 12 instructs the dictionary selection and combination unit 14 to combine the previously stored dictionaries of two genres, “zoo.” And “amusement park.” According to this command, the dictionary selection / coupling unit 14 takes out the facility dictionaries of the zoo and the amusement park in Osaka from the voice recognition dictionary storage unit 13 and combines them. For the N + 1th conversation hierarchy Create a speech recognition dictionary.
[0071]
Next, the dialogue control unit 12 instructs the response voice output unit 15 to output a message prompting the user to speak the name of the facility. In response to this command, the response voice output unit 15 selects a message “Please tell us the name of the facility” from the response voice storage unit 16 and presents it to the user. When the user who hears this message utters the name of the amusement park “OO park.” Which is vaguely stored as the zoo, and inputs it to the voice interaction device 40, the input voice is recognized by the voice recognition unit 11. Is selected, "○ park." Is selected as the recognition result, and the search target is determined.
[0072]
Next, the dialogue control unit 12 instructs the response voice output unit 15 to present the determined search target “OO park.” To the user. In response to this command, the response voice output unit 15 combines the content stored in the response voice storage unit 16 with "OO park." To create a message "display a map of OO park." To the user.
[0073]
With the above operation, a map such as a destination to be searched can be displayed on the display screen of the navigation device.
[0074]
As described above, in the fourth embodiment, the provision of the dictionary selection / combination unit 14, the ambiguous expression word determination unit 38, and the similar concept selection unit 48 allows the user to know, for example, only vaguely about the type of facility. Even if you could not determine the next instruction because you answered the question during the dialogue, such as "Kana zoo", create a speech recognition dictionary that combines the amusement park dictionaries of facilities similar to the zoo By performing the recognition, the search target can be determined without interrupting the flow of the voice dialogue. Therefore, even if the user gets an ambiguous response without knowing the correct answer, the dialog can be continued without interruption and the object can be achieved.
[0075]
Next, FIG. 19 and FIG. 20 are views showing a fifth embodiment of the voice interaction apparatus of the present invention. Since the fifth embodiment is configured substantially in the same manner as the first embodiment described above, the same components are denoted by the same reference numerals using the drawings, and the characteristic portions will be described. explain.
[0076]
First, the device configuration will be described. In FIG. 19, the voice interaction device 50 includes a voice recognition unit 11, a dialogue control unit 12, a dictionary storage unit 13, a dictionary selection connection unit 14, a response voice output unit 15, and a response voice storage unit 16. In addition, a multiple result determination unit 58 is provided instead of the unknown expression word dictionary 17 and the unknown expression word determination unit 18 in the first embodiment, and the multiple result determination unit 58 It is determined whether there are a plurality of speech recognition results in response to the inquiry.
[0077]
Next, the processing operation of the voice dialogue apparatus of the present invention will be described with reference to the flowchart of the dialogue shown in FIG. Here, the purpose is to search for the XX golf course in Chiba Prefecture, but the memory of the location of the XX golf course is ambiguous and the user recognizes that it is in Tokyo or Chiba Prefecture. An example will be described.
[0078]
First, in the same manner as in the first embodiment described above, a voice dialogue is started by a user's instruction, and the user responds to the message "What is your use?" While saying "Facility search.", A speech recognition dictionary of a search genre as shown in FIG. 4 is created, and a message "Please tell us the type of facility" is presented to the user. When the user utters “Golf course” and inputs it to the voice interaction device 50, the input voice is recognized by the voice recognition unit 11, and “Golf course” is selected as a genre of search.
[0079]
Then, the dialog control unit 12 instructs the dictionary selection connection unit 14 to create a dictionary composed of words representing prefecture names in order to narrow down the location of the golf course. In response to this command, the dictionary selection / combination unit 14 creates a speech recognition dictionary composed of prefecture names as shown in FIG. 9 from the speech recognition dictionary storage unit 13.
[0080]
Next, the dialogue control unit 12 instructs the response voice output unit 15 to output a message prompting the user to utter the name of the prefecture where the golf course is located. In response to this command, the response voice output unit 15 selects a message "Please talk about the name of the prefecture where the golf course is located" from the response voice storage unit 16 and presents it to the user. The user who hears this message utters “Tokyo or Chiba Prefecture” because the name of the prefecture where the golf course is located is ambiguous and inputs it to the voice interaction device 50, so that the input voice is recognized by the voice recognition unit 11. And "Tokyo or Chiba Prefecture" is obtained as a recognition result.
[0081]
Next, the dialog control unit 12 outputs this result to the multiple result determination unit 58. The multiple result determination unit 58 receives this result, and determines that two words of Tokyo and Chiba are included in the recognition result as words representing the prefecture name. The result is output to the dialog control unit 12. In response to this determination result, the dialogue control unit 12 instructs the dictionary selection and combination unit 14 to create a dictionary in which the dictionary of the golf course in Tokyo and the dictionary of the golf course in Chiba are combined. In response to this command, the dictionary selection / combination unit 14 extracts the dictionary of the golf course in Tokyo and the dictionary of the golf course in Chiba from the speech recognition dictionary storage unit 13 and combines them to create a speech recognition dictionary.
[0082]
Next, the dialog control unit 12 instructs the response voice output unit 15 to output a message prompting the user to utter the name of the golf course. In response to this command, the response voice output unit 15 selects a message “Please tell us the name of the golf course” from the response voice storage unit 16 and presents it to the user. When the user who hears this message utters the name of the golf course “XX golf course.” And inputs it to the voice interaction device 50, the input voice is recognized by the voice recognition unit 11, and as a recognition result. , "XX golf course." Is selected, and the search target is determined.
[0083]
Next, the dialogue control unit 12 instructs the response voice output unit 15 to present the determined search target “XX golf course.” To the user. In response to this command, the response voice output unit 15 combines the content stored in the response voice storage unit 16 with “XX golf course.” And displays a message “Display a map of golf course.” Create and present to the user.
[0084]
With the above operation, a map such as a destination to be searched can be displayed on the display screen of the navigation device.
[0085]
As described above, in the fifth embodiment, by providing the dictionary selecting / combining unit 14 and the multiple-result determining unit 58, the user does not need to know the name of the prefecture where the golf course is located. Even if the answer to the question during the dialogue was answered and the next instruction could not be determined, a dictionary for the golf course in Tokyo and a dictionary for the golf course in Chiba was taken out and combined to create a speech recognition dictionary. By performing voice recognition, a search target can be determined without interrupting the flow of voice dialogue. Therefore, even if the user gets an ambiguous response without knowing the correct answer, the dialog can be continued without interruption and the object can be achieved.
[0086]
Next, FIGS. 21 and 22 are views showing a sixth embodiment of the voice interaction apparatus of the present invention. Since the sixth embodiment is configured substantially in the same manner as the above-described fifth embodiment, the same reference numerals are given to the same configurations using the drawings, and the characteristic portions will be described. explain.
[0087]
First, the device configuration will be described. 21, the voice interaction device 60 includes a voice recognition unit 11, a dialog control unit 12, a dictionary storage unit 13, a response voice output unit 15, a response voice storage unit 16, and a multiple result determination unit 58. In addition, the dictionary selection unit 24 employed in the second embodiment is employed in place of the dictionary selection / combination unit 14 in the fifth embodiment. In addition, an input storage unit 61 is provided. I have.
[0088]
Here, the voice recognition unit 11 recognizes the voice input by the user and simultaneously outputs the result of acoustic analysis of the input voice to the input storage unit 61. The input storage unit 61 An input voice output from the voice recognition unit 11 or an acoustic analysis result of the input voice is stored.
[0089]
The dictionary selecting unit 24 creates one of the speech recognition dictionaries used by the speech recognition unit 11 by selecting one of the speech recognition dictionaries stored in the dictionary storage unit 13 according to a command from the dialog control unit 12.
[0090]
Next, the processing operation of the speech dialogue apparatus of the present invention will be described with reference to the flowchart of the dialogue shown in FIG.
[0091]
First, in the same manner as in the fifth embodiment described above, a voice dialogue is started by a user's instruction, and in response to a message "What is your need?" The user utters "Facility search.", And then utters "Golf course" in the search genre shown in FIG. 4 in response to the message "Please tell us the type of facility." Since it is "golf course.", A speech recognition dictionary of a search genre as shown in FIG. 9 is created and a message "Please speak the name of the prefecture where the golf course is located" is presented to the user. , The user , In the Nth conversation hierarchy, Since the name of the prefecture where the golf course is located is ambiguous, by saying "Tokyo or Chiba Prefecture" and inputting it to the voice interaction device 50, the input voice is recognized by the voice recognition unit 11, and as a recognition result. , "Tokyo or Chiba Prefecture."
[0092]
Then, the multiple result determination unit 58 determines that the recognition result by the voice recognition unit 11 includes two words representing the name of the prefecture, Tokyo and Chiba. First instructs the dictionary selection unit 24 to select and create a dictionary for a golf course in Tokyo. In response to this command, the dictionary selection unit 24 extracts the dictionary of the golf course in Tokyo from the speech recognition dictionary storage unit 13 For the N + 1th conversation hierarchy Create a speech recognition dictionary.
[0093]
Next, the dialog control unit 12 instructs the response voice output unit 15 to output a message prompting the user to utter the name of the golf course. In response to this command, the response voice output unit 15 selects a message “Please tell us the name of the golf course” from the response voice storage unit 16 and presents it to the user.
[0094]
Next, the dialogue control unit 12 instructs the speech recognition unit 11 to execute speech recognition using the dictionary created by the dictionary selection unit 24. When the user who has heard the previous message utters the name of the golf course “XX golf course.” And inputs it to the voice interaction device 50, the input voice is recognized by the voice recognition unit 11, and the recognition result is obtained. At the same time as "XX golf course." Is obtained, and the input of "OO golf course." Spoken by the user is input in the form of an input voice or in the form of the result of acoustic analysis of the input voice. It is output to the unit 61 and stored.
[0095]
Next, the dialog control unit 12 instructs the response voice output unit 15 to present the recognition result “XX golf course.” To the user. In response to this command, the response voice output unit 15 combines the content stored in the response voice storage unit 16 and “XX golf course.” To create and use the message “XX golf course?” To the person.
[0096]
Next, the dialog control unit 12 instructs the speech recognition unit 11 to execute speech recognition using the dictionary created by the dictionary selection unit 24. On the other hand, when the user utters “No” because the result is not the expected result and inputs it to the voice interaction device 50, the input voice is recognized by the voice recognition unit 11, and the recognition result is “No”. . "Is obtained.
[0097]
Next, the dialogue control unit 12 instructs the dictionary selection unit 24 to create a dictionary of the golf course in Chiba Prefecture because the previous selection dialogue in Tokyo was incorrect. In response to this command, the dictionary selecting unit 24 extracts the dictionary of the golf course in Chiba from the voice recognition dictionary storage unit 13 and creates a voice recognition dictionary.
[0098]
Next, the dialog control unit 12 instructs the voice recognition unit 11 to take out the previous input from the input storage unit 61 and execute voice recognition using the dictionary created by the dictionary selection unit 24, "XX golf course." Is obtained as a recognition result.
[0099]
Next, the dialog control unit 12 instructs the response sound output unit 15 to present “XX golf course.” To the user. In response to this command, the response voice output unit 15 combines the content stored in the response voice storage unit 16 with “XX golf course.” And displays a message “Display a map of golf course.” Create and present to the user.
[0100]
With the above operation, a map such as a destination to be searched can be displayed on the display screen of the navigation device.
[0101]
As described above, in the sixth embodiment, by providing the dictionary selecting unit 24, the multiple-result determining unit 58, and the input storing unit 61, the user can easily find out, for example, the name of the prefecture where the golf course is located. Or Chiba prefecture. ”Even if the answer to the question during the dialogue could not be determined, the results of speech recognition in the Tokyo Metropolitan golf course dictionary and the Chiba prefecture golf course dictionary By sequentially obtaining the speech recognition results and presenting the recognition results, the search target can be determined without interrupting the flow of the speech dialogue. Therefore, even when the user gets an ambiguous response without knowing the correct answer, the dialog can be continued without interruption, and the object can be achieved.
[0102]
【The invention's effect】
As described above, according to the present invention, when the content of the voice of the user whose voice has been recognized is a content in which the next instruction cannot be determined, a voice recognition dictionary effective for continuing the conversation is stored in the dictionary storage unit. Since the preparation is made, the dialogue can be continued by outputting the question voice and the response voice. Therefore, it is possible to provide a speech dialogue device having an excellent effect that a dialogue can be continued without interruption even if a user cannot respond or only an ambiguous answer can be achieved, and an object can be achieved.
[Brief description of the drawings]
FIG. 1 is a related block diagram illustrating a schematic overall configuration of a first embodiment of a voice interaction device according to the present invention;
FIG. 2 is a dialog flow diagram illustrating a processing operation according to the first embodiment of the voice interaction device of the present invention;
FIG. 3 is a conceptual diagram showing a speech recognition dictionary used by the first embodiment of the speech dialogue device of the present invention.
FIG. 4 is a conceptual diagram showing a speech recognition dictionary used by the first embodiment of the speech dialogue device of the present invention.
FIG. 5 is a conceptual diagram showing a speech recognition dictionary used by the first embodiment of the speech dialogue device of the present invention.
FIG. 6 is a conceptual diagram showing a speech recognition dictionary used by the first embodiment of the speech dialogue device of the present invention.
FIG. 7 is a related block diagram showing a schematic overall configuration of a second embodiment of the voice interaction device of the present invention.
FIG. 8 is a dialog flow diagram illustrating a processing operation according to a second embodiment of the voice interaction device of the present invention.
FIG. 9 is a conceptual diagram showing a speech recognition dictionary used by a second embodiment of the speech dialogue device of the present invention.
FIG. 10 is a dialog flow chart for explaining a processing operation showing another embodiment of the second embodiment of the voice dialog device of the present invention;
FIG. 11 is a related block diagram showing a schematic overall configuration of a third embodiment of the voice interaction device of the present invention.
FIG. 12 is a dialog flow diagram illustrating a processing operation according to a third embodiment of the voice dialog device of the present invention;
FIG. 13 is a conceptual diagram showing a speech recognition dictionary used by a third embodiment of the speech dialogue device of the present invention.
FIG. 14 is a conceptual diagram showing a speech recognition dictionary used by a third embodiment of the speech dialogue device of the present invention.
FIG. 15 is a related block diagram illustrating a schematic overall configuration of a fourth embodiment of the voice interaction device of the present invention.
FIG. 16 is a dialog flow diagram illustrating a processing operation according to a fourth embodiment of the voice dialog device of the present invention;
FIG. 17 is a conceptual diagram showing a speech recognition dictionary used in a fourth embodiment of the speech dialogue device of the present invention.
FIG. 18 is a conceptual diagram showing a speech recognition dictionary used in a fourth embodiment of the speech dialogue device of the present invention.
FIG. 19 is a related block diagram illustrating a schematic overall configuration of a fifth embodiment of the voice interaction device of the present invention.
FIG. 20 is a dialog flow chart for explaining the processing operation of the voice interaction device according to the fifth embodiment of the present invention;
FIG. 21 is a related block diagram illustrating a schematic overall configuration of a sixth embodiment of the voice interaction device of the present invention.
FIG. 22 is a dialog flowchart for explaining a processing operation according to a sixth embodiment of the voice dialog device of the present invention;
FIG. 23 is a dialog flow diagram for explaining a processing operation according to the related art.
[Explanation of symbols]
10-60 voice dialogue device
11 Voice Recognition Unit
12 Dialogue control unit
13 Voice recognition dictionary storage
14 Dictionary selection unit
15 Response voice output unit
16 Response voice storage
17 Unknown expression dictionary
18 Unknown expression word judgment unit
24 Dictionary Selector
37 Ambiguous Expression Word Dictionary
38 Ambiguous expression word judgment unit
39 Distance calculator
47 Concept Dictionary Table
48 Similarity concept selector
58 Multiple Result Determination Unit
61 Input storage unit

Claims (2)

Voice recognition means for recognizing words corresponding to the utterance,
Response voice output means for outputting a response voice for urging the user to utter;
Dictionary storage means for storing a speech recognition dictionary that is distinguished for each dialog hierarchy and has relevance between the hierarchies;
Dictionary preparation means for preparing a speech recognition dictionary used for a dialogue among the speech recognition dictionaries stored in the dictionary storage means;
When the word recognized by the speech recognition means in the Nth conversation hierarchy is a combination of an ambiguous expression word and another word that is not an ambiguous expression word, the speech recognition dictionary used in the (N + 1) th interaction hierarchy is The speech recognition dictionary is controlled to prepare a speech recognition dictionary corresponding to another word and a word having a predetermined relationship with the other word, and a response voice for shifting to the (N + 1) th conversation hierarchy is output. And a dialogue control means for controlling the response voice output means . Voice recognition means for recognizing words corresponding to the utterance,
Response voice output means for outputting a response voice for urging the user to utter;
Dictionary storage means for storing a speech recognition dictionary that is distinguished for each dialog hierarchy and has relevance between the hierarchies;
Dictionary preparation means for preparing a speech recognition dictionary used for conversation among speech recognition dictionaries stored in the dictionary storage means,
When there are a plurality of words recognized by the voice recognition means in the Nth conversation hierarchy, the response speech output means is controlled to output a response speech for shifting to the (N + 1) th conversation hierarchy. Preparing a voice recognition dictionary corresponding to one of the plurality of words as a voice recognition dictionary used in the (N + 1) th conversation hierarchy, and performing voice recognition using the prepared voice recognition dictionary by the voice recognition means. A dialogue control unit for controlling the dictionary preparation unit so as to newly prepare a speech recognition dictionary corresponding to another word when the operation fails.

Images