JP2003302995A - Updating method of speech recognition grammar, information processor, and computer program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To efficiently and easily update a speech recognition grammar, which is stored in a user terminal (a client side information processor) that conducts information retrieval using speech recognition, into a speech recognition gram mar stored in a server side. <P>SOLUTION: In an information search system which is constituted of a user terminal UT and a server SV, inputted user's speech is recognized based on a first speech recognition grammar stored in the terminal UT and unknown words included in the voice are detected. When an unknown word is detected in the operation, the server SV generates difference information of two kinds of speech recognition grammars based on the version information of the first speech recognition grammar and the version information of a second speech recognition grammar stored in the server. The terminal UT updates the first speech recognition grammar into the second speech recognition grammar employing the difference information generated by the server SV. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the field of information retrieval processing for retrieving desired information using voice.

[0002]

2. Description of the Related Art Conventionally, a user terminal (client) is connected via a communication line such as the Internet to a server in which various kinds of information are stored in advance, and a user who uses the connected server is connected. Services for searching for desired information from terminals have become widespread, and in recent years,
There is a demand to perform such information retrieval using the voice input at the user terminal.

[0003] In the case of performing information retrieval using a server in response to a voice input at a user terminal, a conventional system generally follows the procedure described below.

FIG. 16 is a flow chart showing an information search process of a user terminal in an information search system using voice input.

In the figure, first, step S1001.
At, the version information of the speech recognition grammar held in the user terminal is acquired. Next, step S1002.
In, the version information of the voice recognition grammar acquired in step S1001 is transmitted to the server via a communication line such as the Internet.

Next, in step S1003, the difference information between the voice recognition grammar necessary for properly searching the information on the server and the voice recognition grammar held in the terminal itself is received. Then, in step S1004, the speech recognition grammar held in the terminal itself and the step S10
In step 03, the voice recognition grammar in the terminal is reconstructed based on the voice recognition grammar difference information received from the server.

Next, in step S1005, the voice uttered by the user is input, and in step S1006.
Then, the input voice is recognized using the voice recognition grammar reconstructed in step S1004.

Then, in step S1007, the recognition result in step S1006 is transmitted to the server as a keyword for retrieval.

In step S1008, step S10
The information corresponding to the keyword transmitted in 07 is received from the server, the received information is output to the own terminal in step S1009, and the process returns to step S1001.

Next, the processing on the server side corresponding to the operation of the user terminal will be described with reference to FIG.

FIG. 4 is a flowchart showing the control processing of the server that provides information.

In FIG. 1, first, in step S801, the process waits until an event such as communication or user input occurs. If an event is detected, the process moves to step S802.

In step S802, step S801
It is determined whether or not the event detected at is an event indicating the reception of version information of the voice recognition grammar, and if it is the reception of grammar version information, the process proceeds to step S803, and if not, the process proceeds to step S806. Transfer.

In step S803, the version of the recognition grammar is received from the user terminal, and in step S804,
Version of the received recognition grammar, step S8 described below
Difference information of the recognition grammar is created using the recognition grammar created in 07 and the recognition grammar difference creation information created in step S808.

Next, in step S805, the difference information of the recognition grammar created in step S804 is transmitted to the user terminal, and the process returns to step S801.

In step S806, step S801
Events detected by are added information to be searched.
It is determined whether it is a database update such as deletion, and if it is a database update, step S807.
If not, the process proceeds to step S809.

In step S807, the speech recognition grammar necessary for searching the information to be searched is created. Next, in step S808, information necessary for creating a difference in recognition grammar is created. The information necessary for creating the difference information of the recognition grammar includes the version information of the recognition grammar and the backup of the recognition grammar corresponding to each version.

In step S809, step S801
It is determined whether or not the event detected in step S6 is the reception of a keyword. If the event is the reception of a keyword, step S81
If not, the process proceeds to step S801.

In step S810, the keyword is received from the user terminal. Next, in step S811, the database is searched for data corresponding to the keyword received in step S810. Next, step S
In 812, the data retrieved in step S811 is transmitted to the user terminal, and the process proceeds to step S801.

[0020]

However, the above conventional example has the following problems.

That is, in order to properly perform the voice recognition on the user terminal, the voice recognition grammar held in the user terminal needs to be an appropriate version, but the user terminal itself uses the voice recognition grammar. It is difficult to determine when to update.

Ideally, when the recognition grammar is created in step S807, the voice recognition grammar on the user terminal may be updated, but it is not always said that the server and the user terminal are connected by a communication line. Not limited to this, it is difficult to synchronize with the timing of creating the recognition grammar on the server.

In many cases, the communication line capacity connecting the server and the user terminal is small, and the system configuration for transmitting the difference to the user terminal each time the voice recognition grammar is updated on the server is not rational. In particular, in the case of changing data that is rarely searched, there is little need to update the voice recognition grammar of the user terminal.

Therefore, according to the present invention, the voice recognition grammar stored in the user terminal (the information processing apparatus on the client side) that performs information retrieval using voice recognition is replaced by the voice recognition grammar stored in the server side. The purpose is to update efficiently and easily.

[0025]

In order to achieve the above object, a method for updating a speech recognition grammar according to the present invention is characterized by the following configuration.

That is, in a method of updating a voice recognition grammar in an information retrieval system composed of a user terminal and a server, the input voice of the user is converted into a first voice recognition grammar stored in the user terminal. A voice recognition step of recognizing based on the voice recognition step and detecting an unknown word included in the voice, and version information of the first voice recognition grammar when the unknown word is detected in the voice recognition step, and the server An update step of updating the first speech recognition grammar to the second speech recognition grammar at the user terminal when the version information of the stored second speech recognition grammar is different. .

Alternatively, in a method of updating a voice recognition grammar in an information retrieval system composed of a user terminal and a server, the inputted voice of the user is converted into a first voice recognition grammar stored in the user terminal. A voice recognition step of recognizing based on the voice recognition step and detecting an unknown word included in the voice, and version information of the first voice recognition grammar when the unknown word is detected in the voice recognition step, and the server A difference generation step of generating difference information between the first speech recognition grammar and the second speech recognition grammar in the server according to the stored version information of the second speech recognition grammar, and a difference generation step. Updating the first speech recognition grammar to the second speech recognition grammar at the user terminal using the generated difference information. And butterflies.

In a preferred embodiment, the difference generation step includes the nth version of the speech recognition grammar and the (n-
1) Difference information between the speech recognition grammar of the n-th version and the speech recognition grammar of the (n−k) version is created by creating difference information with the version of the speech recognition grammar for all k (<n). Should be generated.

Further, for example, in the voice recognition step, when the unknown word is detected, the voice may be re-recognized based on the second voice recognition grammar acquired in the updating step.

Alternatively, there is provided a method of updating a voice recognition grammar in a user terminal connectable to a server for information retrieval for achieving the same object, wherein an inputted user voice is stored in the user terminal. A voice recognition step of recognizing based on a first voice recognition grammar and detecting an unknown word included in the voice, and a version of the first voice recognition grammar when an unknown word is detected in the voice recognition step. The first speech recognition grammar is transmitted to the server and the second speech recognition grammar stored in the server is converted into the second speech recognition grammar by using the difference information acquired from the server in response to the transmission of the version information. And an updating step of updating.

It is to be noted that the same object is to implement a method for updating a speech recognition grammar having each of the above-mentioned configurations by a computer having a communication function, and a computer-readable storage medium storing the program code. Also achieved by.

[0032]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of an information retrieval system according to the present invention will be described in detail below with reference to the drawings.

[First Embodiment] FIG. 1 is a system configuration diagram of an information retrieval system according to the first embodiment.

In the figure, the user terminal UT and the server S
The V is connected via a telephone line or a communication line such as the Internet in a so-called server / client environment by a general procedure, and is connected for bidirectional communication. In the present embodiment, the user inputs a voice using the voice input device H305 connected to the user terminal UT, and the server SV
The information stored in the external storage device H12 connected to can be searched. The information retrieved by the server SV is notified to the user by voice and / or display at the user terminal UT.

FIG. 2 is a block diagram showing a hardware configuration of a user terminal and a server which constitute the information search system in the first embodiment.

In the user terminal UT shown in FIG.
Is a central processing unit (C
PU), and performs calculation according to the procedure of information search processing described later. H2 is an output device such as a speaker or a display that presents information to the user by voice and / or display, and presents the user with voice recognition results, database search results, and the like.

H3 is an input device through which the user gives an operation instruction to the user terminal UT or inputs information. The input device H3 is a general-purpose input device such as a touch panel and a keyboard, and a voice input device H3 such as a recognition button H301, a selection button H302, a direction button H303, a dial H304, and a microphone for inputting voice.
Including 05.

H4 is a storage device such as a disk device or a non-volatile memory, and stores a software program for information retrieval processing described later, a data communication program with the server SV, a voice recognition grammar and version information thereof. There is. H5 is a read-only storage device (ROM) in which fixed data such as a boot program is stored.
H6 is a storage device such as a RAM that holds temporary information, and holds temporary data and various flags.

H7 is a modem, LAN card, P
A data communication device such as an HS card, which is used for communication with the server SV. The components of the user terminal UT described above are connected by an internal bus.

The user terminal UT shown in FIG. 1 has, for example, a so-called PDA (personal digital assistant) shape.
A personal computer having a wired and / or wireless communication function can be adopted as the user terminal UT in the information search system described in this embodiment and the second embodiment described later.

Next, in the server SV shown in FIG. 2, H
Reference numeral 8 denotes a central processing unit (CPU) that performs processing such as numerical calculation and control, and performs calculation according to a procedure of control processing described later. H9 is an output device such as a display for presenting information to the user.

H10 is an input device through which a user gives an operation instruction to the device or inputs information.
H11 is a data communication device such as a modem, a LAN card, and a PHS card, and is a client (user terminal UT).
Used to communicate with.

H12 is a storage device such as a disk device or a non-volatile memory, and holds a database, voice recognition grammar, recognition grammar difference creation information, and the like. H13 is a read-only storage device (ROM) in which a boot program, fixed data for terminal identification, and the like are stored.

H14 is a storage device such as a RAM for holding temporary information, which holds temporary data and various flags. The components of the server SV described above are connected by an internal bus.

Next, the processing executed by the user terminal UT and the server SV in the above-mentioned information retrieval system will be described.

FIG. 3 is a flow chart showing the information retrieval process performed by the user terminal UT in the first embodiment, and shows the procedure of the process performed by the central processing unit (CPU) H1.

In the figure, first, in step S701, the voice uttered by the user is input via the voice input device H305, and in step S702, the input voice is input to the external storage device in voice recognition step S2. Recognize using the voice recognition grammar held in H4.

In step S703, step S701
Unknown word in the speech input in step S7
At 02, a word not included in the speech recognition grammar used for recognition is detected.

Next, in step S704, step S
In 703, it is determined whether an unknown word is detected,
As a result, if an unknown word is detected, step S708.
Otherwise, to step S705.

In step S705, step S702
Is transmitted to the server SV using the communication device H7 as a keyword, and in step S706,
In response to the transmission of the keyword, the information corresponding to the keyword is received from the server SV. Then, the received information is output to the output device H2 in step S707, and then the process returns to step S701.

On the other hand, if an unknown word is detected in step S704, in step S708 the speech recognition grammar currently stored in the external storage device H4 (first
(Speech recognition grammar) is determined, and if it is determined that the current grammar is updated as a result, the process proceeds to step S709; otherwise, step S705.
Proceed to.

Here, whether or not the voice recognition grammar should be updated is determined depending on whether the voice recognition grammar has already been updated or whether the grammar update is prohibited by the user.
That is, when the same unknown word is detected again even though the voice recognition grammar has already been updated, or when the user's setting instruction for prohibiting the update of the voice recognition grammar is given in advance, etc. The speech recognition grammar stored in the external storage device H4 is not updated.

In step S709, the external storage device H4
The version information of the currently recognized speech recognition grammar (first speech recognition grammar) is read, and in step S710, the read version information is transmitted to the server SV using the communication device H7.

In step S711, step S710
In response to the version information being transmitted at, the difference information from the voice recognition grammar currently held in the external storage device H4 is received from the server SV. Next, step S712.
Then, the speech recognition grammar currently held in the external storage device H4 and the difference information received from the server SV are used to reconstruct the speech recognition grammar into the same speech recognition grammar as the second speech recognition grammar held by the server SV. Then, the process returns to step S702. Here, as a procedure for reconstructing the speech recognition grammar, a general procedure can be adopted at present, and therefore the description in this embodiment will be omitted.

Next, as for the processing of the server SV corresponding to the operation of the user terminal UT, the control processing shown in FIG. 4 described above can be adopted, and the description thereof will be omitted.

According to the information retrieval system of this embodiment, the timing of updating the voice recognition grammar on the user terminal UT can be automatically determined, and the timing of updating the voice recognition grammar can be determined. , User terminal UT
Only when the user utters a word other than the speech recognition grammar currently stored in. As a result, the number of communications between the user terminal UT and the server SV can be reduced as compared with the system configuration in which the information search processing described with reference to FIG. 16 is executed on the client side. Even if the capacity is relatively small, the voice recognition grammar can be updated efficiently.

<Modification of First Embodiment> First Embodiment
In the embodiment of step S711 and step S711.
Although the difference of the recognition grammar is transmitted and received in 805, the entire recognition grammar may be transmitted and received in some cases. For example, if the recognition grammar version received by the server SV in step S803 is very old, the entire recognition grammar is transmitted. At that time, by transmitting together with the information indicating that the transmission content is the entire recognition grammar, the user terminal U
In step S712, T configures the recognition grammar to be totally rewritable. With this system configuration,
This is preferable because the amount of information such as the backup of the recognition grammar created in step S808 can be suppressed.

[Second Embodiment] Next, a second embodiment based on the information retrieval system according to the first embodiment will be described. In the following description, description of the system configuration (FIGS. 1 and 2) that can adopt the same configuration as that of the first embodiment is omitted, and the system configuration described above is executed by the user terminal UT and the server SV. The processing performed will be described.

FIG. 5 to FIG. 9 are flowcharts showing the information retrieval process performed by the user terminal UT in the second embodiment, and show the procedure of the process performed by the central processing unit (CPU) H1. In the present embodiment, the software program of the information search process is described as an event-driven program, and interrupt information such as a user's input operation from the input device H3 and synchronization information with a subroutine operating in parallel, etc. Is published as an event.

In the figure, first, in step S1, a new event is acquired, and in step S2, it is determined whether or not the event acquired in step S1 is the depression of the recognition button H301 by the user. If it is, the process proceeds to step S3. If it is not, the recognition button H301 is not pressed, step S4.
Transfer processing to.

In step S3, the voice input device H305
After starting the voice recognition of the user input via, the process proceeds to step S1. Here, the voice recognition in the present embodiment operates asynchronously with the processes shown in the flowcharts of FIGS. 5 to 9, and general voice recognition is executed as the voice is input from the voice input device H305. It When the voice recognition result is confirmed in this step, an event indicating the end of voice recognition is issued.

In step S4, it is determined whether or not the event acquired in step S1 is a release operation of the recognition button H301 by the user, and the recognition button H301 is detected.
If the recognition button H301 is not released, the process proceeds to step S5. If the recognition button H301 is not released, the process proceeds to step S6.

In step S5, the execution of voice recognition started in step S3 is stopped, and the process proceeds to step S1.

In step S6, it is determined whether or not the event acquired in step S1 is an event indicating the end of voice recognition. If it is the end of voice recognition, the process proceeds to step S101. The processing moves to S7.

In step S7, it is determined whether the event acquired in step S1 is caused by the user operating the dial H304. If the event is the dial H304 operation, the process proceeds to step S201, and the dial H3 is operated.
If the operation is not 04, the process proceeds to step S8.

In step S8, it is determined whether or not the event acquired in step S1 is the pressing of the selection button H302 by the user. If the event is the pressing of the selection button H302, the process proceeds to step S301 and the selection button H302 If not pressed, the process proceeds to step S1.

In step S101, the result obtained by voice recognition is acquired, and in step S102, the input voice corresponding to the recognition result obtained in step S101 is acquired.

Next, in step S103, step S
It is determined whether the speech recognition result obtained in 101 includes an unknown word, and if the unknown word is included, step S10.
If the unknown word is not included, the process proceeds to step S112.

In step S104, the external storage device H4
It is determined whether or not to update the voice recognition grammar (first voice recognition grammar) currently stored in. If the grammar needs to be updated, the process proceeds to step S105. If the grammar needs not to be updated, the process proceeds to step S112. Transfer. For example, when there is an input instruction from the user not to update the voice recognition grammar, it is determined that the grammar need not be updated.

In step S105, data that means "request for recognition grammar" is transmitted to the server SV, and in step S106, the version information of the grammar currently used for speech recognition in the terminal itself is acquired.

Next, in step S107, step S
The version information of the voice recognition grammar obtained in 106 is transmitted to the server SV, and in step S108, in response to the transmission of the version information, the server SV transmits the voice recognition grammar stored in the server (second voice recognition). Grammar) and the difference information is received.

Next, in step S109, step S
A new speech recognition grammar that is the same as the second speech recognition grammar is reconstructed based on the difference information obtained in step 108 and the speech recognition grammar currently used for speech recognition in the terminal itself.

Next, in step S110, the version information of the new grammar is received from the server SV.

Next, in step S111, step S
The voice recognition grammar newly constructed in step S109 is applied to the voice acquired in step 102 to perform voice recognition, and the process proceeds to step S112.

In step S112, step S101
Alternatively, the speech recognition result obtained in step S111 is
Output to the output device H2.

Next, in step S113, step S
In step 112, the first rank recognition result in the list displayed (output) on the output device H2 (for example, “Kinkakuji” highlighted in the display example of FIG. 12) is selected in step S1.
Transfer processing to.

In step S201, dial H304
The direction of the user's input operation performed with respect to is determined, and if the operation direction is upward, the process moves to step S202, and if the operation direction is not upward, step S203.
Transfer processing to.

In step S202, the recognition result of the selection target by the user is moved to the recognition result in the upper direction. That is, if the nth recognition result is selected, (n
-1) th recognition result is selected, and then step S
The process is transferred to 1.

In step S203, the recognition result of the selection target by the user is moved downward by one. That is, if the nth recognition result is selected, (n +
The 1) th recognition result is selected, and then the process proceeds to step S1.

In step S301, the selected recognition result is acquired, and in step S302, data meaning "request for image data" is transmitted to the server SV. Then, in step S303, the maximum number of images received by the terminal is transmitted to the server SV.

Next, in step S304, step S
The contents of the selection acquired as the recognition result in 301 are stored in the server S.
Send to V.

Next, in step S305, step S
The number of images corresponding to the recognition result transmitted in 304 is received from the server SV, and in step S306, step S3
It is determined whether the number of images received in 05 is not 0, and the number of images is 0.
If so, the process proceeds to step S307. If the number of images is not 0, the process proceeds to step S308.

In step S307, the output device H2 warns the user that the image data does not exist, and the process proceeds to step S1. An example of the warning that the image data does not exist is shown in FIG.

In step S308, step S305
It is determined whether the number of images received in step S303 is larger than the maximum number of received images transmitted to the server SV. If the number of images is larger than the maximum number of received images, the process proceeds to step S309, and the number of images is maximum received image. If it is less than the number, the process proceeds to step S313.

In step S309, step S301
The user is warned that the number of images that can be acquired from the server SV, which corresponds to the selection content acquired as the recognition result, is larger than the maximum number of received images. FIG. 14 shows an example of a warning that the image data is larger than the maximum number of received images.

Next, in step S310, step S
A list of unique keywords that uniquely identify the image corresponding to the selection content acquired as the recognition result in step 301 is received from the server SV, and in step S311, step S3 is performed.
The unique keyword list received in 10 is output to the output device H2. Then, in step S312, the first keyword is selected from the output unique keywords, and the process proceeds to step S1.

In step S313, the first image data of the images corresponding to the selected recognition result is stored in the server S.
Received from V, and in step S401, step S30
It is determined whether the number of images received in 5 is greater than 1, and if the number of images is greater than 1, the process proceeds to step S402, and if the number of images is not greater than 1, the process proceeds to step S403.

In step S402, the second and subsequent image data are received from the server SV asynchronously with the processing of the flowchart shown in FIG. After starting the asynchronous reception, the process proceeds to step S403.

In step S403, the screen of the output device H2 is switched to the image display screen, and in step S404, the image data received in step S313 is displayed, for example, as illustrated in FIG.

Next, in step S405, an event is acquired. In step S406, it is determined whether the event acquired in step S405 is an operation of the dial H304 by the user. If the event is an operation of the dial H304, a step S411 is performed. Move the process to, dial H
If it is not the operation of 304, the process proceeds to step S407.

In step S407, step S405
The event acquired in step is the recognition button H301 by the user
It is determined whether or not it is due to the pressing of the
If the button is pressed, the process proceeds to step S408, and if the recognition button H301 is not operated, the process proceeds to step S405.

In step S408, step S402
Asynchronous image data reception started in step S409 is stopped, and in step S409, the screen of the output device H2 is switched to the original screen as illustrated in FIG. 12, for example.

Next, in step S410, step S
As in the case of 3, voice recognition is started, and the process proceeds to step S1 to acquire an event.

In step S411, the target image is changed according to the operation of the dial H304 by the user. n
If the th image is the current target image, if the operation of the dial H304 is in the upward direction, the (n-1) th image is the target image, and if the operation of the dial H304 is in the downward direction, the (n + 1) th image. The image is the target image.

Next, in step S412, step S4
At 411, it is determined whether the target image changed has already been received by the server SV and exists in the terminal itself. If the target image has already been received, the process proceeds to step S414, and if the target image has not already been received, step S414.
Processing is moved to 413.

In step S413, step S411
The process waits until the target image changed in step S402 is received by the asynchronous image data reception started in step S402, and when the target image changed in step S411 is received, the process proceeds to step S414. Step S414
Then, the target image changed in step S411 is displayed, and the process proceeds to step S405.

Next, the operation of the server SV in this embodiment will be described.

10 and 11 are flowcharts showing the control processing performed by the server SV in the second embodiment, showing the procedure of the processing performed by the central processing unit (CPU) H8. In the present embodiment, the control processing software program is described as an event-driven program, and interrupts such as input from the input device H10 and data reception from the client (user terminal UT) are events. Issued as.

In the figure, first, in step S501, a new event is acquired. In step S502, it is determined whether the event acquired in step S501 is data reception from the user terminal UT, and the user terminal UT is detected.
If the data is received from the user terminal UT, the process proceeds to step S506. If the data is not received from the user terminal UT, the process proceeds to step S503.

In step S503, step S501
It is determined whether the event acquired in step 5 is a notification of database update in the external storage device H12 such as addition / deletion of image data. If the event is a database update notification, the process proceeds to step S504, and a database update notification is issued. If not, the process proceeds to step S501.

In step S504, a speech recognition grammar (second speech recognition grammar) adapted to the updated database is created. For example, when the data is added, the corresponding word is added, and when the data is deleted, the word is deleted.

Next, in step S505, step S
Information necessary to create a difference between the grammar created in 504 and the past grammar is created. Specifically, all n
For (n is a natural number), the grammar created for the nth time (that is, the speech recognition grammar of the nth version) and the grammar created for the (n-1) th time (that is, the speech recognition of the (n-1) th version) You can make a grammar) difference. This makes it possible to create a difference between the nth grammar and the (n−k) th grammar for all k (<n). After the difference creation information of the recognition grammar is created in this step, step S
The process is moved to 501.

In step S506, the data received from the user terminal UT, which is the user terminal UT, is the "image data request" in step S302 (FIG. 8) described above.
If the data means "request for image data", the process moves to step S601. If the data does not mean "request for image data", the process moves to step S507.

In step S507, it is determined whether the received data is the data that means "request for recognition grammar" in step S105 (FIG. 6) described above, and if the data means "request for recognition grammar". Step S50
If the data does not mean "request for recognition grammar", the process proceeds to step S501.

In step S508, the version information of the recognition grammar held in the user terminal UT is sent to the server S.
Receive to V.

Next, in step S509, step S
Difference information of the recognition grammar is created by the version information of the speech recognition grammar (first speech recognition grammar) received in 509 and the recognition grammar difference creation information created in step S505. Then, in step S510, step S509
The difference information of the recognition grammar created in step 1 is transmitted to the user terminal UT. Furthermore, in step S511, the version information of the current speech recognition grammar (second speech recognition grammar) is transmitted to the user terminal UT, and the process proceeds to step S501.

In step S601, the upper limit of the number of image data to be sent to the user terminal UT is received, and in step S602, the database search keyword is received from the client (user terminal UT).

Next, in step S603, the data corresponding to the received keyword is searched from the database in the external storage device H12, and in step S604, the number of data obtained as a result of the search is transmitted to the user terminal UT. To do.

Next, in step S605, it is determined whether or not the number of data obtained as the search result is larger than the upper limit of the number of image data sent to the user terminal UT received in step S601, and the data obtained as the search result. Is greater than the upper limit of the number of image data to be sent to the user terminal UT received in step S601, the process is moved to step S607, and the number of data obtained as the search result is the number of user terminals received in step S601. If the number of image data to be sent to the UT is less than the upper limit, step S60
Processing is transferred to 6.

In step S606, step S603 is performed.
The data obtained as a result of the search in 1 is transmitted to the user terminal UT, and the process proceeds to step S501.

In step S607, step S603 is performed.
Create a keyword list to identify the data obtained as a result of the search in. Examples of unique keywords are keywords that are manually assigned in consideration of uniqueness,
For example, there are connections of all keywords assigned to data.
Further, it is not always necessary to uniquely identify, and any keyword that can be narrowed down to the maximum number of transmitted images or less may be used.

Then, in step S608, the unique keyword list created in step S607 is transmitted to the user terminal UT, and the process proceeds to step S501.

<Modification of Second Embodiment> Second Embodiment
In the embodiment of the present invention, in the user terminal UT, step S
After the unique keyword list is displayed on the output device H2 at 311, the speech recognition grammar may be a recognition grammar composed of only unique keywords. In this case, after executing step S409 for displaying the initial screen, all You can switch to a grammar that recognizes keywords.

In the second embodiment described above, a plurality of image data stored in the database in the external storage device H12 is described as a search target. However, the search target of the above information search system is an image. The present invention is not limited to this, and may be applied to voice data or any other search target.

In the second embodiment described above, in the user terminal UT, the operation of the upward direction button H303 can be regarded as the upward operation of the dial H304, and the operation of the downward direction button H303 can be regarded as the downward operation of the dial H304. For example, the operation with the direction button H303 is also possible.

According to the information retrieval system according to the present embodiment and its modification, the timing for updating the voice recognition grammar on the user terminal UT can be automatically determined, and the voice recognition grammar is updated. The timing is
Only when the user utters a word other than the speech recognition grammar currently stored in the user terminal UT. As a result, the number of communications between the user terminal UT and the server SV can be reduced as compared with the system configuration in which the information search processing described with reference to FIG. 16 is executed on the client side. Even if the capacity is relatively small, the voice recognition grammar can be updated efficiently.

[0117]

Other Embodiments It is an object of the present invention to provide a communication function for operating a storage medium (or recording medium) recording a software program code for realizing the functions of the above-described embodiments as the above-mentioned user terminal UT. By supplying the information processing apparatus such as a personal computer and a server computer that operates as the server SV, and the computer (or CPU or MPU) of those systems or apparatuses reads and executes the program code stored in the storage medium. To be achieved. In this case, the program code itself read from the storage medium realizes the function of the above-described embodiment, and is acquired as a computer program product via the storage medium storing the program code, the electric communication line, or the like. The program code constitutes the present invention.

Further, by executing the program code read by the computer, not only the functions of the above-described embodiments are realized, but also the operating system (OS) running on the computer is executed based on the instruction of the program code. ) And the like perform part or all of the actual processing, and the processing realizes the functions of the above-described embodiments.

Furthermore, after the program code read from the storage medium is written in the memory provided in the function expansion card inserted in the computer or the function expansion unit connected to the computer, based on the instruction of the program code. It also includes a case where the CPU or the like included in the function expansion card or the function expansion unit performs a part or all of the actual processing and the processing realizes the functions of the above-described embodiments.

[0120]

According to the present invention described above, the voice recognition grammar stored in the user terminal (the information processing device on the client side) that retrieves information using voice recognition is
The voice recognition grammar stored on the server side can be updated efficiently and easily.

[Brief description of drawings]

FIG. 1 is a system configuration diagram of an information search system according to a first embodiment.

FIG. 2 is a block diagram showing a hardware configuration of a user terminal and a server that form the information search system in the first embodiment.

FIG. 3 is a flowchart showing an information search process performed by a user terminal UT in the first embodiment.

FIG. 4 is a flowchart showing a control process of a server that provides information.

FIG. 5 is a flowchart showing an information search process performed by a user terminal UT in the second embodiment.

FIG. 6 is a flowchart showing an information search process performed by a user terminal UT in the second embodiment.

FIG. 7 is a flowchart showing an information search process performed by a user terminal UT in the second embodiment.

FIG. 8 is a flowchart showing an information search process performed by a user terminal UT in the second embodiment.

FIG. 9 is a flowchart showing an information search process performed by the user terminal UT in the second embodiment.

FIG. 10 is a flowchart showing a control process performed by the server SV in the second embodiment.

FIG. 11 is a flowchart showing a control process performed by a server SV in the second embodiment.

FIG. 12 is a diagram illustrating a display screen when a plurality of recognition results are displayed in a list on the output device H2 of the user terminal UT.

FIG. 13 is a diagram illustrating a display screen that displays an image corresponding to a recognition result on the output device H2 of the user terminal UT.

FIG. 14 shows that the number of image data on the server SV is the user terminal U.
When the number of received images is larger than the maximum number of received images of T, the user terminal UT
FIG. 7 is a diagram exemplifying a guidance screen for warning the user in FIG.

FIG. 15 is a diagram exemplifying a guidance screen for warning the user at the user terminal UT that the corresponding image data does not exist on the server SV.

FIG. 16 is a flowchart showing an information search process of a user terminal in the information search system using voice input.

Front page continuation (51) Int.Cl. ⁷ Identification code FI theme code (reference) G10L 15/00 G10L 3/00 537J 15/18 551A 15/20 571K 15/28 531P

Claims (8)

[Claims] 1. A method for updating a voice recognition grammar in an information retrieval system including a user terminal and a server, wherein an input user voice is converted into a first voice recognition grammar stored in the user terminal. A voice recognition step of recognizing based on the voice recognition step, and detecting an unknown word included in the voice; and when an unknown word is detected in the voice recognition step, version information of the first voice recognition grammar and the server. An updating step of updating the first speech recognition grammar to the second speech recognition grammar at the user terminal when the version information of the stored second speech recognition grammar is different;
A method for updating a speech recognition grammar, comprising: 2. A method of updating a voice recognition grammar in an information retrieval system comprising a user terminal and a server, wherein an inputted voice of a user is converted into a first voice recognition grammar stored in the user terminal. A voice recognition step of recognizing based on the voice recognition step, and detecting an unknown word included in the voice; and when an unknown word is detected in the voice recognition step, version information of the first voice recognition grammar and the server. A difference generation step of generating difference information between the first speech recognition grammar and the second speech recognition grammar in the server according to the stored version information of the second speech recognition grammar; Using the generated difference information,
And a step of updating the first speech recognition grammar with the second speech recognition grammar in the user terminal. 3. In the difference generating step, difference information between the speech recognition grammar of the nth version and the speech recognition grammar of the (n−1) th version is created for all k (<n), The method for updating a speech recognition grammar according to claim 2, wherein difference information between the nth version of the speech recognition grammar and the (n−k) th version of the speech recognition grammar is generated. 4. In the voice recognition step, when the unknown word is detected, the voice is re-recognized based on the second voice recognition grammar acquired in the updating step. The method for updating the speech recognition grammar according to claim 2 or claim 3. 5. A method of updating a voice recognition grammar in a user terminal connectable to an information retrieval server, wherein the input user's voice is based on a first voice recognition grammar stored in the user terminal. Voice recognition step of detecting the unknown word included in the voice, and transmitting the version information of the first voice recognition grammar to the server when the unknown word is detected in the voice recognition step. At the same time, an updating step of updating the first speech recognition grammar to the second speech recognition grammar stored in the server using the difference information acquired from the server in response to the transmission of the version information. A method for updating a speech recognition grammar, which comprises: 6. An information processing device connectable to a server for information search, which recognizes an input user's voice based on a first voice recognition grammar stored in the information processing device,
A voice recognition unit that detects an unknown word included in the voice; and when the voice recognition unit detects an unknown word, the version information of the first voice recognition grammar is transmitted to the server, and the version information of the version information is transmitted. Update means for updating the first speech recognition grammar to the second speech recognition grammar stored in the server using the difference information acquired from the server in response to the transmission. Information processing equipment. 7. A computer program product, wherein the method for updating a voice recognition grammar according to claim 5 gives an operation instruction which can be realized by a computer having a communication function. 8. A computer program as the information processing apparatus according to claim 6, which gives instructions to operate a computer having a communication function.