JP2009087172A - Method of creating program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To form a program composed of a plurality of scenario execution programs of which an execution order can be changed. <P>SOLUTION: Content of the programs A, B are decomposed into scenarios 1-3, 4-6, and the program A, B are assembled from them. The scenarios 1-6 are formed by arranging a basic scenario which has common parts in voice recognition or voice synthesis. A program C can be created by decomposing into scenarios 8-10, and arranging the basic scenario. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a method for creating a program using voice recognition.

  Conventionally, when developing a speech recognition system, it has been necessary to develop from scratch, from a speech recognition engine, to linkage with a database, a speech dictionary, and a speech synthesis engine. That is, in a program using speech recognition, since it is integrated as one file and is a process specific to the system, for example, as shown in FIG. 11A, the execution order of the program A and the program B To change the order, the program had to be recreated to change the order. Further, for example, as shown in FIG. 11B, when trying to create a program C that allows the number of groups and the number of people to be determined at the same time, the program C is created as a completely different program from the programs A and B described above. There was a need to do. In other words, it cannot be diverted as a program for other systems.

By the way, in a system having a communication function between a plurality of systems including a system that controls a service and a system that executes the service, a plurality of scenarios for each service are stored as service scenarios, and the service scenario is customized by customizing the service scenario. Have been proposed (see, for example, Patent Document 1).
JP2003-32367A (paragraph 0037)

  However, the technique described in Patent Document 1 can be applied only to a system having a communication function between a plurality of systems including a system that controls a service and a system that executes the service.

  In view of this, the present invention assumes that the contents of each program are made up of a set of a plurality of scenario execution programs whose execution order can be changed. It is an object of the present invention to provide a method for creating a program that enables the above.

  The invention of claim 1 is a method of creating a program for creating a program using speech recognition or speech synthesis, wherein the program is set as an aggregate of a plurality of scenario execution programs whose execution order can be changed. The scenario execution program is produced by customizing a basic scenario execution program having a common part in the speech of speech recognition or speech synthesis.

  In this way, the program is made up of a collection of a plurality of scenario execution programs whose execution order can be changed, and each scenario execution program is created by customizing the basic scenario execution program. Therefore, it is possible to easily change the contents of the program using speech recognition or speech synthesis.

  According to a second aspect of the present invention, in the program creation method of the first aspect, the basic scenario execution program can select a logic such as a line type, a voice recognition engine, a voice synthesis engine, a voice dictionary, and the like. It is desirable to do.

  In this way, the logic can be selected corresponding to each scenario execution program.

  According to a third aspect of the present invention, in the method for creating a program according to the first or second aspect, the basic scenario execution program is a plurality of types, and a request scenario execution program, a confirmation scenario execution program, and a selection request are requested. A scenario execution program and an operation execution scenario execution program can be included.

  In this way, it is possible to cope with the creation of various patterns of programs by combining the question, confirmation, selection request, and operation execution programs.

  In the present invention, the program is made up of a collection of a plurality of scenario execution programs whose execution order can be changed, and each scenario execution program is created by customizing a basic scenario execution program. Therefore, it is possible to easily change the contents of the program using voice recognition.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a system configuration diagram showing a schematic configuration of a voice recognition system using a program using voice recognition according to the present invention.

  As shown in FIG. 1, a speech recognition system 1 using a speech recognition development tool is provided in a specific company, and a speech server 2 that executes a speech recognition program and a customer database server 3 that stores customer data. And a private branch exchange 5 (PBX) for carrying out a conversation with the mobile phone 6 or the fixed phone 7 through the telephone line 4. An operator 8 can communicate with a user of a mobile phone 6 or a fixed phone 7 through a telephone line 4 via a private branch exchange 5.

  The voice recognition system 1 is linked to a maintenance company 10 through the Internet 9 for maintenance and inspection.

  When such a speech recognition system 1 is used, for example, in a call center, the scenario flow can be as shown in FIG. In other words, when a customer receives a message, he / she speaks first as "Customer service desk. Please choose one from maintenance, new contract, and maintenance procedure." On the other hand, when the customer answers “Maintenance.”, He / she utters “Are you sure you want to introduce conservation?”

  On the other hand, when the customer answers “Yes!”, He / she utters “I will connect to the operator”. On the other hand, when the customer replies “Maintenance contract”, he / she utters “Are you sure about the maintenance procedure?”.

  On the other hand, when the customer answers “Yes!”, He / she utters, “Please say one of cancellation, account change, name change, address change.”

  On the other hand, when the customer answers “cancel”, he / she utters “Are you sure you want to cancel?”.

  On the other hand, when the customer answers “Yes!”, He / she utters “Please tell me your insurance number”.

  On the other hand, when the customer replies “1234!”, He / she utters “Are you sure that the contract number is 1234? Taro Yamada?”

  On the other hand, when the customer answers “Yes!”, The customer responds by saying “Thank you for using the cancellation document.” The process ends.

  In addition, when used in the call center described above, the customer may answer “Maintenance” or “Yes!”, And when the answer is recognized, a confirmation sound such as “Pee” may be emitted. Of course.

  Further, although a specific description is omitted, the scenario flow can be as shown in FIG. 3 when used for taxi dispatch. In this case, the content of the customer's utterance is played as audio as it is, and the data is transmitted to the dispatch system database at the time when the reception is completed, so the dispatcher issues a taxi dispatch instruction here.

  Next, a method for creating a program using specific voice recognition used in the voice server 2 will be described.

  The entire program is created as an aggregate of a plurality of scenario execution programs whose execution order can be changed. Each scenario execution program is created by customizing a basic scenario execution program. That is, each scenario execution program can change the word (speech), change the utterance content / order, set the usage database, select the usage dictionary, and the like.

  The basic software (scenario writer) used to create a scenario is used to create an approximate processing flow on a GUI basis. A plurality of prepared scenario execution programs are compiled to create a scenario list.

In the basic scenario execution program, various types of logic such as a line type, a voice recognition engine, a voice dictionary, and the like can be selected. Each scenario has scenario name, initial setting, line type, data type, voice recognition engine, barge-in function, voice dictionary, input data, external data execution file, output data, utterance execution file, condition, scenario transfer judgment, extension number transfer Items such as determination, utterance content, speech synthesis engine, new scenario addition, scenario copy, saved data, conversation execution file, confirmation sound, etc. are items that can be selected or changed for each scenario.

  Also, as shown in FIG. 2 and FIG. 3, each scenario has a scenario execution program that asks “Please say ...”, a scenario execution program that confirms “... Execution of the scenario execution program that requires the selection of "... select one ...", "I will connect you to the operator" and "Please wait for a while as I will turn around the taxi soon." Since the basic scenario execution programs corresponding to the respective scenario execution programs are prepared, all the scenario execution programs can be created by arranging (changing) them.

  If the execution order of the respective scenario execution programs is determined and integrated, a program for sequentially executing a series of operations is completed.

  That is, as shown in FIG. 4A, the contents of the programs A and B are decomposed into scenarios 1 to 3 and 4 to 6, and the programs A and B are produced as an aggregate of them. In addition, each scenario 1 to 6 is prepared by arranging the basic scenarios, so that it can be easily changed and added and deleted easily. Therefore, as shown in FIG. 4 (b), the program C can also be created by arranging the basic scenarios by disassembling the scenarios 8 to 10, and there is no need to newly recreate them as in the prior art. .

  Subsequently, a procedure for creating a new scenario list on a personal computer using the basic software will be specifically described.

  First, as shown in FIG. 5, when a screen for creating a new scenario is opened, a new scenario frame 11 for adding a scenario is created. Here, for example, file1 is the scenario number. Note that the scenario number can be changed. For example, the scenario number can be changed by clicking file1.

  When the scenario frame 11 is right-clicked, a property list 11a is displayed. Each item in the property list is an item that can be selected or changed for each scenario described above. Each of these items is sequentially registered according to the scenario to be created. A connection branch point 11 b is displayed on the scenario frame 11.

  For example, in the case of performing “initial setting” as to whether or not the scenario is executed first, when the initial setting in the property list is clicked, “Yes” and “No” are displayed as shown in FIG. Usually, since it is not the scenario to be executed first, it is “None”. However, in the case of the scenario to be executed first, “Yes” is set as shown in FIG.

  In the next item “line type”, a digital line or an analog line is selected. When an analog line is selected, it is as shown in FIG. Note that it is also possible to support an IP phone.

  In the item “data type”, a voice or a telephone line is selected.

  In the item “Speech recognition engine”, select Nuance 8.5 (manufactured by Nuance Communication) or others.

  In the item "speech synthesis engine", select FineSpeech (Fujitsu), recorded WAV file or other.

  In the item “barge-in function”, “ON” is selected when jumping to the next scenario even during the utterance, and “OFF” is selected when not jumping.

  In the item “voice dictionary”, as shown in FIG. 6C, another window is displayed, and a voice dictionary to be recognized is set there. Note that a specified file can be selected and executed for each scenario.

  Although not specifically illustrated, another window is displayed for the subsequent items “input data” and “external data execution file”, and settings are made there.

  In the case of “with scenario transfer determination”, another scenario connection branch point 11 c is displayed on the scenario frame 11.

  After all the scenarios are created in this way, the entire scenario is completed by connecting the branch points 11b and 11c of each scenario frame 11 in the scenario design document, as shown in FIG. When scenario script generation is executed after the whole is completed, the scenario execution program is completed.

  In the above embodiment, the voice recognition system using the voice recognition development tool has been described. However, as shown in FIG. 8, the voice recognition system can be configured using the voice APS service.

  In this case, as shown in FIG. 8, as the IDC 21 (Internet Database Connecter), a voice server 2 that executes a voice recognition program, a customer database server 3 that stores customer data, and a mobile phone through a telephone line 4. 6 and a private branch exchange 5 (PBX) for carrying out a conversation with the fixed telephone 7. An operator 8 can communicate with a user of a mobile phone 6 or a fixed phone 7 through a telephone line 4 via a private branch exchange 5.

  The IDC 21 is linked to the maintenance company personal computer 10, the user personal computer 13, and the operator 8 through the Internet 9.

  As an example of the scenario in this case, an example applied to the voice version ASP service (golf reservation) will be described.

  When there is a call from a mobile phone or landline, first speak at the IDC, "Here is the XX golf reservation reception center. Please choose one from reservation, empty information, confirmation."

  In response, the customer answers “reservation”.

  In response to this, he utters "Reservation begins. Please tell us the reservation date and start time." Here, if it is not a reservation, if you say “No”, you will return to the previous one.

  In response to this, the customer responds “Starts around 8 o'clock on October 1”.

  In response to this, he says, “October 1st, start at 8 o'clock. Please tell us how many people will play and the number of groups.” At this time, if the date or time is different, say “No” to go back one step.

  In response, the customer replied “4 people in one set”.

  In response to this, he says, “The number of groups to play is one group and the number of players is four. Then, we will confirm the reservation, so please wait for a while.” Here, if the number of groups or the number of players to play is different, saying “No” will return to the previous one.

  In response, the customer responds “Yes!”.

  In response to this, he says, "You can start at 9:32, but will you make a reservation?" Here, if the start time is not the desired time, saying “No” will announce a different time.

  In response, the customer answers “cancellation”.

  In response to this, he said, “October 1 at 9:32 in the IN start, 4 people took one set. Would you like to return to the first menu?”

  On the other hand, if the customer answers “No”, it ends.

  If the customer says “I want to talk”, it is forwarded to the operator.

  FIG. 10 shows an example in which the voice version ASP service is applied to a restaurant reservation in addition to a golf reservation. In this case as well, in the case of a golf reservation, if it is not a reservation, if you say “No”, you will return to the previous one, and if the date or time is different, say “No” , We will return to the previous one and will be transferred to the staff by the customer's utterance "I want to talk".

  The present invention includes the above-mentioned call center reception work, taxi outside vehicle dispatch reception, golf course reservation reception, restaurant / restaurant reservation reception, hotel / inn reservation reservation, school / classroom reservation reception, hospital / clinic reservation Order reservation reception, delivery / delivery order reception, beauty salon reservation, mail order company product order reception, nursing care reservation reception, confirmation of various charges, search of various shops (phone numbers, etc.), fortune-telling, dictionary, proverb It can also be applied to timetables, fares, distances, price search services, schedule management services, emergency contact information telephone number services, document / catalog billing agency services, questionnaire services, product explanations, and guidance services.

1 is a system configuration diagram showing a schematic configuration of a speech recognition system using a program using speech recognition according to the present invention. It is explanatory drawing which shows an example of a scenario. It is explanatory drawing which shows an example of a scenario. (A) (b) is a figure which shows the relationship between a program and a scenario, respectively. It is a figure which shows the relationship between a scenario frame and a scenario property. (A), (b), and (c) are explanatory diagrams of initial setting, line type selection, and voice dictionary selection, respectively. It is explanatory drawing of the scenario design document which integrates a scenario. It is a system configuration diagram in the case of configuring a speech recognition system using a speech APS service. It is explanatory drawing which shows an example of a scenario. It is explanatory drawing which shows an example of a scenario. It is explanatory drawing of the program of a prior art example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Voice recognition system 2 Voice server 3 Customer database server 4 Telephone line 5 Private branch exchange 6 Cellular phone 7 Fixed telephone 8 Operator 9 Internet 11 Scenario frame 21 IDC

Claims (3)

A method of creating a program for creating a program using speech recognition or speech synthesis,
The program as a set of a plurality of scenario execution programs whose execution order can be changed,
A method for producing a program, characterized in that each scenario execution program is created by customizing a basic scenario execution program having a common part in the speech of speech recognition or speech synthesis.   2. The program creation method according to claim 1, wherein the basic scenario execution program can select various types of logic such as a line type, a speech recognition engine, a speech synthesis engine, and a speech dictionary.   The basic scenario execution program includes a plurality of types of a scenario execution program for a question, a scenario execution program for confirmation, a scenario execution program for requesting selection, and a scenario execution program for executing an operation. A method for producing the program according to 1 or 2.

Images