JP2000181475A - Voice answering device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To inform a user of the reliability of an answer sent back by an answering system to a question inputted by the user and also to adaptively vary the detailedness of the answer. SOLUTION: An input analysis block 2 makes a retrieval block 5 to retrieve data according to information inputted from an input block 1, an answer text generation block 4 generates an answer sentence from the retrieved data, and a voice synthesis block 3 outputs a synthesized voice of the answer sentence. The answer text generation block 4 generates the answer sentence on which the reliability of the answer is reflected, varies the abstraction degree of the expression of the corresponding answer sentence according to the accuracy of the retrieved data, and also varies the abstraction degree of the answer sentence at a request for accuracy estimated from input information and a history of a history management block 9.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system for responding to a question or the like input from, for example, a user, and more particularly to a voice response device for responding to a natural language voice.

[0002]

2. Description of the Related Art Conventionally, there is a system that responds to a question or the like input from a user or the like. As an example of the response system, there is known a voice response device that responds to a question or the like from a user or the like by voice in a natural language.

[0003] There is also a speech synthesis system and a speech synthesis method that change the response voice according to the content of the response and the source of the information.

[0004]

Incidentally, when a response is returned from a response system to a question input by a user or the like, for example, there is a case where it is desired to inform the user of the reliability of the response. However, in the conventional voice response device,
Many tone and sound quality are constant. In this case, the user cannot understand the reliability unless he / she listens to the contents of the response.

[0005] Further, for example, even if a system that changes the response voice according to the content of the response and the source of the information is used,
Not enough to convey the reliability of the response. For example, if the user is uneasy to understand the question, the response is low in reliability. On the other hand, even if the detailed part of the information used for the response is uncertain, it is considered that a sufficient response can be made unless the user's request requires details.

Further, adaptively changing the level of detail of a response is also an important point in voice response. That is, for example, if the explanation is made in more detail than necessary by voice output, it cannot be said that the response is appropriate. One method of responding in such a case is disclosed in Japanese Unexamined Patent Application Publication No. H8-1376.
As disclosed in Japanese Patent Publication No. 98, there may be a method in which the format and history of the question from the user are reflected, and only the portions deemed necessary are described, and the others are omitted. Also,
As another method, the degree of detail may be changed depending on the degree of abstraction when converting data into a response sentence.
For example, when notifying the date of a certain event, is it to be informed of "month, season", or about a specific "date and time"?

However, in the conventional voice response system, when converting data into a response sentence, many data items are put into a prepared response sentence template. , It cannot support abstraction for simple response.

The present invention has been made in view of such circumstances, and when a response is returned from a response system to a question input by a user or the like, the user is informed of the reliability of the response. Is possible, and
It is an object of the present invention to provide a voice response device capable of adaptively changing the degree of detail of a response.

[0009]

According to the present invention, there is provided a voice response apparatus comprising: input means for inputting information; search means for performing data search based on the input information; and a response for generating a response sentence from the searched data. When generating a response sentence based on the retrieved data, a response sentence that reflects the reliability of the response is generated. By generating, the above-mentioned problem is solved.

[0010] The voice response apparatus according to the present invention comprises: input means for inputting information; search means for performing data search based on the input information; and response sentence generation means for generating a response sentence from the searched data. Means for converting a response sentence into a synthesized sound and outputting the synthesized sound, and when generating a response sentence based on the retrieved data, generating a response sentence based on the accuracy of the retrieved data. The above-described problem is solved by changing the degree of abstraction.

[0011] The voice response apparatus according to the present invention comprises: input means for inputting information; search means for performing data search based on the input information; and response sentence generation means for generating a response sentence from the searched data. Generating a response sentence based on the retrieved data, including history management means for managing a history of past inputs and responses to the input, and synthetic sound conversion output means for converting the response sentence into a synthesized sound and outputting the synthesized sound In doing so, the above-described problem is solved by changing the degree of abstraction of the expression of the response sentence according to the request for accuracy estimated from the input information and the history.

That is, according to the present invention, when returning a response to a question input from a user or the like, the tone, voice, etc. of the response sentence are determined according to, for example, the type, certainty, detail, and amount of data. By changing the response, it is possible to inform the user about the reliability of the response, and it is also possible to adaptively change the level of detail of the response, and to further simplify the response sentence according to the points specified by the user. Output.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described with reference to the drawings.

FIG. 1 shows a configuration example of a voice response system according to an embodiment to which the voice response device of the present invention is applied.

First, as an example of the voice response system configuration of the first embodiment, an information search system that searches for information corresponding to a question from a user, generates a voice from the searched information, and outputs the voice will be described as an example.

In FIG. 1, an input block 1 is an input conversion means for receiving an input from a user, converting the input into a numerical value or a symbol, or converting the input into a text.
Here, the input from the user is, for example, a keyword, a command, a question, and the like. Examples of the input include an input by a user operating a button, a keyboard, a touch panel, and the like, and an input by a voice to a microphone. Conceivable.

The input analysis block 2 receives the processing result of the input block 1, interprets the user input, determines what kind of information is to be used as a keyword, and determines data to be searched. Inform 5 In addition, the input analysis block 2 determines how detailed the data obtained as a result of the search in the search block 5 is to be output as voice, and notifies the response text generation block 4 of the determination result.

The history management block 9 has a role of supplementing information for making a decision in the input analysis block 2 when the information is lacking. It manages the history of the user's past input and the response output made to it. The history management block 9 sends an answer to the inquiry about the missing information from the input analysis block 2.

The retrieval block 5 retrieves desired data from the information stored in the database block 6 based on the information from the input analysis block 2. It should be noted that the data obtained as a result of the search includes information other than what the user wants to know in the end and more detailed information than necessary. The data of the search result is sent to the response text generation block 4 as it is.

The database block 6 is a place where information to be searched is recorded. The information stored in the database block 6 includes information prepared by the system such as a map and a dictionary, as well as information edited and added later by the user, such as a schedule table and a telephone directory. It also includes information obtained via a network, such as traffic information, news, and weather forecasts.

The response text generation block 4 generates a response text based on the data received from the search block 5. That is, response text generation block 4
Then, for example, based on information from the input analysis block 2,
Data items that are not required by the user are cut or the level of abstraction of the expression is determined. Further, the response text generation block 4 extracts words from the text generation data storage block 8 and connects them to generate a text, and determines the quality and loudness of voice when the text is output as voice.

The text generation data storage block 8 has a set of words corresponding to each element of the data and a set of endings for expressing the characteristics of the output information.
The correspondence between data and words is one-to-many, for example, “October 5
The expressions "Sun", "Early October" and "This Autumn" are all 1
Data. Which one to select depends on the accuracy of the data and the degree of abstraction of the expression determined in the speech text generation block 4. In addition, the ending is selectively used so that the characteristics of the output information can be intuitively conveyed to the user.
For example, when the reliability of the response is high, the suffix “is” is used, and when the reliability is low, the suffix is used.

The speech synthesis block 3 receives the text and its output method from the response text generation block 4. The speech synthesizing block 3 acquires necessary data from the speech synthesizing data storage block 7 based on the data, and connects them to generate synthesized speech data. Then, the synthesized speech data of the response sentence generated in the speech synthesis block 3 is finally sent as an audio signal to the speaker, whereby the synthesized speech of the response sentence is output from the speaker.

The speech synthesis data storage block 7 is provided with a data set such as phonemes for converting text into synthesized speech. The information prepared in the voice synthesizing data storage block 7 is selectively used so that the characteristics of the output information can be intuitively transmitted to the user. That is, in the system of the present embodiment, for example, when the reliability of the response is high, the voice of an adult is used, and when the reliability of the response is low, the voice of a child is used.

Here, factors that determine the reliability of the response include, for example, confidence in the interpretation of the user input, the source of the information used for the response, and the accuracy expected by the user for the response. .

First, an example in which confidence in the interpretation of the user input is a factor that determines the reliability of the response will be specifically described.

For example, if the user inputs a question "What is the time from Tokyo to Shizuoka?"
The details of the means of transportation, such as whether to use a car or a train to travel from Tokyo to Shizuoka, are missing. Here, it is assumed that the input analysis block 2 manages the history of the user input stored in the history management block 9, and from the history of the history management block 9, for example, it is found that the user often moves by car. . In that case,
The system according to the present embodiment does not ask the user about the contents of the means of transportation that are missing in the user's question, but can check the travel time by car for the time being and return a response. However, in this case, it is unknown whether it is correct to return the travel time by car in the response, and the reliability of the response is reduced accordingly. That is, the confidence in the interpretation of the user input indicates, for example, how reliable the interpretation is when the user input is interpreted as described above, and the confidence in the interpretation of the user input indicates the reliability of the response. Is determined.

Next, the source of the information used for the response is
An example serving as an element that determines the reliability of the response will be specifically described.

For example, when calculating the travel time by car from Tokyo to Shizuoka, for example, a method of calculating the travel time by car using a road map or the travel time by car using traffic information There is a method to calculate. When the road map is used, for example, a road distance between two points, a departure point and a destination, is calculated, and the travel distance is calculated by dividing the road distance by an expected hourly speed. When using traffic information,
For example, it is provided between two points of departure and destination,
Check the time required between each measurement point such as traffic volume and traffic jam,
The travel time is obtained by summing them. As described above, assuming that there are two sources of information used for the response, a case where a road map is used and a case where traffic information is used, the traffic information that can take into account real-time traffic volume, traffic congestion information, etc. When used, the reliability of the response is higher. Thus, the reliability of the response is determined by the source of the information used for the response.

Next, an example will be described in which the accuracy that the user expects for the response is a factor that determines the reliability of the response.

For example, it is assumed that the travel time by car from Tokyo to Shizuoka is calculated from a road map, and a calculation result of 2.1 hours is obtained as the travel time. Here, assuming that the accuracy is, for example, only in units of one hour, it is considered that a response of "about two hours" to the question "What is the time from Tokyo to Shizuoka?" However, in response to the question "How long does it take from Tokyo to Shizuoka?"
The minute response is considered unreliable. That is, even if the response is from the same data, the reliability of the response changes depending on the comparison with the user's expectation. Thus, the reliability of the response is determined by the accuracy that the user expects of the response.

On the other hand, factors that determine the degree of abstraction of the data expression include, for example, the accuracy of the data itself and the accuracy that the user expects for a response.

First, an example in which the accuracy of the data itself is an element that determines the degree of abstraction of the data expression will be specifically described.

For example, it is assumed that the travel time by car from Tokyo to Shizuoka is calculated from a road map, and a calculation result of 2.1 hours is obtained as the travel time. In such a case, the way of expressing the response to the user is as follows:
It is assumed that the calculated result of the traveling time, 2.1 hours, is expressed as “2 hours and 6 minutes”, “2 hours a little”, or “about 2 hours”. The expression can be changed according to the error. In this way, the degree of abstraction of the data representation of the response is determined according to the accuracy of the data itself.

Next, an example will be described in which the accuracy that the user expects for the response is an element that determines the degree of abstraction of the data expression.

For example, a user asks "What is the schedule for tomorrow?"
When asked, "I'm playing tennis in the morning." On the other hand, in response to the question "What is your plan for tomorrow morning?", The user is asked to answer specifically "Time is tennis from 9:00 to 11:00." Thus, the degree of abstraction of the data representation is determined by the accuracy that the user expects for the response. Note that the accuracy that the user expects for the response can be estimated not only from the shape of the question input from the user but also from the past history. For example, if a user always asks for the exact time, and asks, "What is your plan for tomorrow?"
Ask to answer, "I play tennis from 9:00 to 11:00."

Hereinafter, the operation of the system according to the first embodiment will be specifically described with reference to FIGS.

Here, it is assumed that when the user inputs a question together with a keyword, a response by voice output is returned from the system of the present embodiment.

First, as an initial setting, a mode is set in which the user operates the input block 1 to obtain information on the road.

Next, as step S1, the user inputs a question "What is the time from Tokyo to Shizuoka?" Here, it is assumed that a voice input is assumed as an interface for inputting a question, and a user speaks into a microphone of the system. The input sound from the microphone is
It is subjected to voice recognition processing in the input block 1 and converted into text data.

The input analysis block 2 analyzes the text data sent from the input block 1 and determines a corresponding search process. That is, the input analysis block 2 has a list of keywords for determining processing, and first searches for a word that matches this keyword.

Here, "What is the time from Tokyo to Shizuoka?"
Find three words that match the keywords: "Tokyo", "Shizuoka", and "Time". At this time, it can be understood from a place name such as “Tokyo” and “Shizuoka” and a keyword “time” that the traffic information is checked, but the transportation means (moving means) cannot be understood as it is. Therefore, the input analysis block 2 makes an inquiry to the history management block 9. When the input analysis block 2 learns from this inquiry that, for example, a car has been mostly selected as a means of transportation in the past, the input analysis block 2 assumes that the means of transportation is a car in step S10 based on this. Thus, it is assumed that the user does not explicitly indicate transportation means (moving means), but the possibility of a car is high from the history, and he is confident in understanding user input.

Next, the input analysis block 2 instructs the search block 5 to “acquire traffic information on the Tomei Expressway” in step S2, and instructs the response text generation block 4 to “search for traffic information on the Tomei Expressway”. Calculates and outputs the required time between Tokyo and Shizuoka from the information obtained. "
Also, the input analysis block 2 determines from the user's question format that it is sufficient to notify the approximate time as a response, and provides the response text generation block 4 with an abstract representation of the minute unit of time. To do so.

The search block 5 accesses the database block 6 in accordance with the instruction from the input analysis block 2 to obtain traffic information. At this time, the search block 5 checks information stored locally in the database block 6 which is a storage unit of the system. If the information of the database block 6 is old or missing, for example, the latest information is sent via a network. Get information. In addition, where traffic information can be obtained from the network, and the operation of the system for obtaining that information,
Assume that it is predetermined and known. The traffic information obtained by the search in the search block 5 is sent to the response text generation block 4.

The response text generation block 4 at this time
Is notified from the input analysis block 2 that “the time required between Tokyo and Shizuoka is calculated and output”. First, at step S3, for example, “Tokyo-shizuoka” is extracted from the traffic information obtained by the search in the search block 5. -Atsugi, Atsugi-Gotemba
The required time of each section of "Gotemba-Shizuoka" is picked up as necessary data, and the total time (215 minutes in this example) is calculated as step S4.

Next, the response text generation block 4 generates a response text. First, the expression of time is determined. In the case of the above example, the minute unit is determined to be an abstract expression according to the instruction from the input analysis block 5. In determining this abstract expression, the response text generation block 4 includes a text generation data storage block 8.
And select an expression such as "half" from among four words such as "(do not say anything)", "slightly", "half" and "weak" as the words of the abstract expression. The time of 215 minutes is "3 and a half hours"
To the expression. In addition, since the calculated time is based on traffic information and has high accuracy, the expression of the ending is changed to a disjunctive "is", and furthermore, a response having sufficient reliability to user input is provided. It is determined that there is a voice, and an instruction relating to text and voice synthesis for outputting a voice with a large and clear tone of an adult voice is output to the voice synthesis block 3.

The speech synthesis block 3 receives a text “3 and a half hours” and an instruction regarding speech synthesis from the response text generation block 4. Thereby, the voice synthesis block 3 generates voice output data of a large and clear tone of an adult voice, "3 and a half hours", as a step S11, and outputs it as a response from the speaker.

The information source for examining "time from Tokyo to Shizuoka" is not limited to traffic information. Although less accurate, it is also possible to estimate the required time by examining the distance from a road map.

In this case, the input analysis block 2 informs the search block 5 as "obtain map information of the Tomei Expressway" at step S6.

At this time, the search block 5 accesses the database block 6 to obtain map information. Information obtained by the search block 5 searching the database block 6 is sent to the response text generation block 4.

At this time, the response text generation block 4 informs the input analysis block 2 that "the required time between Tokyo and Shizuoka is to be output", and proceeds to step S7.
From the map information obtained by the search of the search block 5, for example, "Tokyo-Atsugi", "Atsugi-Gotenba", "Gotemba-
The distance of each section of "Shizuoka" is picked up as necessary data, and the total distance (210 km in this example) is calculated as step S8. Further, in the response text generation block 4, the moving speed of the car is averaged at 100 per hour.
km, the required time of 2.1 hours is roughly estimated from the total distance and the average speed per hour.

Next, the response text generation block 4 generates a response text. Also in this case, first, the expression of time is determined. In the case of the above example, the minute unit is determined to be an abstract expression according to the instruction from the input analysis block 5. In determining the abstract expression, the response text generation block 4 refers to the text generation data storage block 8 and uses the words “(say nothing)”, “slightly” and “slightly” as the words of the abstract expression as described above. Half and weak
For example, the expression "(do not say anything)" is selected from among the four, and thereby the above 2.1 hours is expressed as "2 hours" in step S9. In addition, the calculated time is assumed to be appropriate as an approximate time, and the expression at the end is made to be a disjunctive "is". However, when map information is used, the reliability of the response is lower than when the above-mentioned traffic information is used. Is output to the speech synthesis block 3.

The speech synthesis block 3 receives from the response text generation block 4 the text "2 hours" and an instruction regarding speech synthesis. As a result, the speech synthesis block 3 sets “2 hours” in step S12.
, But generates voice output data of a small voice but outputs it as a response from the speaker.

In the case where the time is calculated from the road map as described above, if the question from the user is not "What is the time from Tokyo to Shizuoka?" The following responses are possible.

In this case, in the input analysis block 2, it is determined that the user wants a response of the expression up to the minute unit.

Thus, the response text generation block 4
Then, based on this, the 2.1 hour obtained by the above calculation is expressed as “2 hours 6 minutes”. However, assuming that the minute unit of the calculation result is unreliable, the "6 minutes" portion in the response is highly likely to be incorrect (lie). Therefore, in the response text generation block 4, since the reliability of the response is not sufficient, "kana" expressing ambiguity is used at the end. Then, the response text generation block 4 outputs, to the voice synthesis block 3, a text for performing voice output in, for example, a child's voice and an instruction regarding voice synthesis.

The speech synthesis block 3 receives from the response text generation block 4 the text "2 hours and 6 minutes" and an instruction regarding speech synthesis. As a result, the voice synthesis block 3 generates the voice output data of the child's voice “2 hours 6 minutes” and outputs it as a response from the speaker.

Further, for example, if a policy is made not to use data representation that cannot guarantee accuracy, the question from the user is "How many minutes from Tokyo to Shizuoka?" Then, the following response can be considered.

In this example, in the input analysis block 2,
It is determined that the user wants a response of the expression up to the minute unit.

Thus, the response text generation block 4
Then, based on this, we will try to express the 2.1 hours obtained by the above calculation as "2 hours 6 minutes". However, here, if the minute unit of the calculation result is unreliable, the "six minutes" portion of the response is likely to be incorrect (lie). Therefore, in the response text generation block 4, the expression is increased in the degree of abstraction, and the expression is determined to be "2 hours" without saying a specific minute unit. Further, the response text generation block 4 appends “approx.” To the end of the response text to reflect that the answer cannot be answered any more.
However, in this case, the response does not satisfy the user's request.
For example, a text for performing voice output with a child's voice and an instruction regarding voice synthesis are output to the voice synthesis block 3.

The speech synthesis block 3 receives from the response text generation block 4 a text “about 2 hours” and an instruction regarding speech synthesis. As a result, the voice synthesis block 3 generates the voice output data of the child's voice “about 2 hours” and outputs it as a response from the speaker.

Next, as a configuration of the voice response system according to the second embodiment of the present invention, a scheduler for managing a user's schedule will be described as an example. Hereinafter, the operation of the system according to the second embodiment will be specifically described with reference to FIGS.

First, as an initial setting, a mode is set in which the user operates the input block 1 to obtain personal information such as schedules and telephone numbers.

Next, as step S31, the user inputs, for example, a question "What is the schedule for tomorrow?" Here, it is assumed that a voice input is assumed as an interface for inputting a question, and a user speaks into a microphone of the system. The voice input from the microphone is subjected to voice recognition processing in the input block 1 and is converted into text data.

The input analysis block 2 analyzes the text data sent from the input block 1 and determines a corresponding search process. That is, the input analysis block 2 has a list of keywords for determining processing, and first searches for a word that matches this keyword.

Here, from the question "What is the plan for tomorrow?", Two words "tomorrow" and "plan" are found as words that match the keyword. Then, the input analysis block 2 determines to search the schedule table from the keyword “planned”, and determines to check tomorrow from the keyword “tomorrow”.

Therefore, in step S32, the input analysis block 2 instructs the search block 5 to “acquire scheduled data of tomorrow”, and the response text generation block 4 specifies “acquired data of tomorrow”. Let me know everything. " However, in this case, the input analysis block 2
Determines that the user wants to know roughly from the form of the user's question, and does not say the detailed time and details of the schedule.

The search block 5 receiving the search instruction from the input analysis block 2 accesses the database block 6 and extracts the schedule of tomorrow as step S33. The information extracted at this time is “9:
00-11: 00 tennis at Shinagawa "," 14: 00-
Shopping at Shinjuku with friends. " Information obtained by the search in the search block 5 is sent to the response text generation block 4.

Next, the response text generation block 4 generates a response text. At this time, the time is first determined for the expression, but as described above, the input analysis block 2
, The response text generation block 4 determines to increase the abstraction level of the expression. For example, referring to the text generation data storage block 8, there are, for example, three classifications of "morning", "day", and "night". Therefore, the response text generation block 4 selects "9: 00-11:""00" is "morning", "14: 00-
(After 14:00) "is determined to correspond to" noon ".
Regarding the contents of the schedule, the place and the partner are omitted, and only the business is said. That is, the response text generation block 4 generates a response text such as “Tennis in the morning” or “Shopping in the afternoon” in step S34.

Next, in the response text generation block 4,
Determine sentence endings and instructions for speech synthesis. As in the second embodiment, in the case of the scheduler, there is no other mistake in the interpretation of the user input, and if the response data is accurate and correct because the user himself has entered the response data, Reliability is sufficient. Therefore, in step S35, the response text generation block 4 uses the disjoint type "is" as the expression of the ending, and issues a speech and voice synthesis instruction for performing voice output in a large and clear tone of an adult voice. Output to 3.

The voice synthesizing block 3 receives from the response text generating block 4 the text “Tennis in the morning and shopping in the afternoon” and instructions on voice synthesis. Thereby, the speech synthesis block 3 performs step S3
As 9, the voice output data of a large and clear tone is generated by the voice of an adult saying, "Tennis in the morning and shopping in the afternoon."

However, when the user asks "What is the schedule for tomorrow?" As in the above example, the date and time and the business may be answered in detail.

For example, in the input analysis block 2, “from the information of the history management block 9,
After that, the date and time are often asked again. " In this case, in the input analysis block 2, the history is prioritized over the expression of the current input, and the response text generation block 4 is instructed to "detail each tomorrow's schedule."

In this case, in the response text generation block 4, a text and voice for performing voice output in a loud and clear tone of an adult voice such as “Tennis from 9 am to 11 am and shopping from 2 pm” An instruction regarding synthesis is output to the speech synthesis block 3.

When the voice synthesis block 3 receives the text and the voice synthesis instruction from the response text generation block 4, the voice synthesis block 3 speaks out loud as an adult saying, "Tennis from 9 am to 11 am, shopping from 2 pm." Generates voice output data with a clear tone and outputs it as a response from the speaker.

On the other hand, for example, let us consider a case where the user operates the input block 1 to specify a mode for obtaining personal information, and then inputs, for example, the question "What is the schedule for tomorrow morning?" Also in this case, as in the above, it is assumed that voice input is assumed as an interface for inputting a question, and the user speaks into the microphone of the system. The voice input from the microphone is subjected to voice recognition processing in the input block 1 and is converted into text data.

In this case, the input analysis block 2 analyzes the text data sent from the input block 1 and determines a corresponding search process. The input analysis block 2 has a list of keywords for deciding the processing, and first searches for a word that matches this keyword.

Here, from the question "What is the plan for tomorrow morning?", Three words "tomorrow", "morning" and "plan" are found as words that match the keyword. Then, the input analysis block 2 determines to “search for the tomorrow's schedule” from the combination of the keywords “tomorrow” and “schedule”. At this time, the input analysis block 2
Instructs the search block 5 to "acquire the data of the tomorrow's schedule" in step S32, and instructs the response text generation block 4 to "inform only the schedule of tomorrow morning". Further, in this case, the input analysis block 2 determines from the format of the user's question that the user wants to know the detailed time,
In addition, try to say more details.

The search block 5 receiving the search instruction from the input analysis block 2 accesses the database block 6 and extracts the schedule of tomorrow as step S33. Information obtained by the search in the search block 5 is sent to the response text generation block 4.

Next, the response text generation block 4 generates a response text. At this time, the response text generation block 4 is informed by the input analysis block 2 that “only the schedule in the morning is output”, so the response text generation block 4 determines “14:
Since the schedule of "00-" is not a schedule in the morning, it is excluded from the output, but the contents of the schedule are all output for time, place, and business. Also, in the response text generation block 4, in order to inform the time as much as possible,
"9: 00-111: 00" is expressed as "from 9:00 to 11".

Next, in the response text generation block 4,
Determine sentence endings and instructions for speech synthesis. In this example, if there is no other mistake in the interpretation of the user input and the data of the response is correct by the user himself, the reliability of the response is sufficient. Therefore, in the response text generation block 4, as a step S38, the expression of the ending is used as a disjunctive type "is", and "from 9:00 to 11
It is tennis in Shinagawa until time. To the voice synthesizing block 3 for outputting a text and voice synthesis for voice output in a large and clear tone of an adult voice.

When the voice synthesis block 3 receives the instruction regarding the text and voice synthesis from the response text generation block 4, in step S 40, the voice of an adult saying “Tennis is played at Shinagawa from 9:00 to 11:00” is clearly pronounced. The tone output data is generated and output as a response from the speaker.

Next, FIG. 4 shows a hardware configuration for realizing the voice response system of the present invention.

In FIG. 4, an input unit 11 corresponds to the input block 1 of FIG.
It has an interface such as a touch panel and a microphone. The user inputs a question or a keyword from the input unit 11.

The CPU 12 controls various parts of the system and executes various program processes. The CPU 12 is responsible for various processes in the input analysis block 2, the speech synthesis block 3, the response text generation block 4, and the like in FIG.

The ROM 13 stores fixed data and fixed programs. The above-described text generation and speech synthesis data and programs for processing are also stored in the ROM 13.

The RAM 14 temporarily stores necessary data during the processing of the program by the CPU 12. The input from the user, the result of text generation and speech synthesis, information obtained by the communication unit 16, and the like are temporarily stored in the RAM.

The auxiliary storage unit 15 stores a flash memory or an E
It is composed of an EPROM and stores data to be always added to data to be added, rewritten, and the like, even during program processing. The personal information of the history management block 9 in FIG.
Data obtained by communication, other data used in the processing of the CPU, and the like are stored in the auxiliary storage unit 15.

The communication unit 16 is used when the system obtains or transmits information via a network. For example, communication such as telephone, the Internet, infrared rays, and radio is performed.

The speaker section 17 outputs a voice such as a voice-synthesized response.

The display section 18 includes a display device such as a liquid crystal panel, and is a so-called GUI (Graphical US).
er Interface) screen, tables, lists, etc.
Display various information.

The auxiliary storage unit 15, the communication unit 16, and the display unit 18 are not necessarily required in the system according to the embodiment of the present invention.

As described above, according to the embodiment of the present invention, the tone, sound quality, and the like when outputting a response sentence to a question from a user by voice are changed according to the reliability of the response. The user can hear the audio output while knowing to some extent whether or not a response matching his / her request has been returned. Further, according to the present embodiment, even if the user simply listens to the audio output, the state of the system can be easily understood, and the next countermeasure (for example, listening properly, entering an insufficient condition, asking another question, etc.) is performed quickly. It is possible. Further, according to the present embodiment, by changing the abstraction of the data expression in the response output, it is possible to indicate the degree of accuracy and to simplify the output.

[0094]

As is apparent from the above description, the voice response apparatus of the present invention performs a data search based on the input information, generates a response sentence from the searched data, and outputs it as a synthesized sound. Generating a response sentence reflecting the reliability of the response, or changing the degree of abstraction of the expression of the response sentence according to the accuracy of the retrieved data;
Alternatively, by changing the degree of abstraction of the expression of the response sentence according to the request for accuracy estimated from the input information and the history, when the response system returns a response to the question input from the user or the like, the response is returned. It is possible to inform the user about the reliability of the response, and it is also possible to adaptively change the level of detail of the response.

That is, according to the present invention, by changing the tone and sound quality when outputting a response sentence according to the reliability of the response, whether the user has returned a response that meets his or her request. It is possible to listen to the audio output while knowing to some extent, and even if only the audio output is heard, the state of the system becomes easy to understand, and the next countermeasure becomes faster. Further, according to the present invention, by changing the degree of abstraction of the data expression in the response output, it is possible to indicate the degree of accuracy and simplify the output.

[Brief description of the drawings]

FIG. 1 is a functional block diagram illustrating a schematic configuration of a voice response system according to an embodiment of the present invention.

FIG. 2 is a flowchart illustrating a flow of an operation example of the system according to the first embodiment.

FIG. 3 is a flowchart illustrating a flow of an operation example of the system according to the second embodiment;

FIG. 4 is a block circuit diagram showing a hardware configuration for realizing the voice response system of the present invention.

[Explanation of symbols]

1 input block, 2 input analysis block, 3 speech synthesis block, 4 response text generation block, 5
Search block, 6 Database block, 7
Speech synthesis data storage block, 8 Text generation data storage block, 9 History management block, 11
Input unit, 12 CPU, 13 ROM, 14 R
AM, 15 auxiliary storage unit, 16 communication unit, 17
Speaker part, 18 display part

Continuing from the front page (72) Inventor Hirofumi Tsunoda 6-35, Kita-Shinagawa, Shinagawa-ku, Tokyo Inside Sony Corporation (72) Inventor Yasuhiko Kato 6-35, Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Stock In-house F term (reference) 5D015 AA05 KK04 5D045 AA07 AB21

Claims (3)

[Claims] An input unit for inputting information; a search unit for performing data search based on the input information; a response sentence generation unit for generating a response sentence from the searched data; And a synthetic sound conversion output means for converting the sound into sound and outputting the sound. The response sentence generating means, when generating a response sentence based on the retrieved data, generates a response sentence reflecting the reliability of the response A voice response device. 2. An input unit for inputting information, a search unit for performing data search based on the input information, a response sentence generation unit for generating a response sentence from the searched data, and a synthesis of the response sentence And a synthetic sound conversion output means for converting the response data into sound. The response sentence generating means, when generating a response sentence based on the retrieved data, responds to the response according to the accuracy of the retrieved data. A voice response device characterized by changing the degree of abstraction of sentence expression. An input unit for inputting information; a search unit for performing data search based on the input information; a response sentence generation unit for generating a response sentence from the searched data; A history management unit that manages a history of responses to the input; and a synthesized sound conversion output unit that converts the response sentence into a synthesized sound and outputs the synthesized sound. The response sentence generation unit includes a response based on the searched data. When generating a sentence, the voice response apparatus changes the degree of abstraction of the expression of the response sentence according to a request for accuracy estimated from the input information and the history.