JP2005242183A - Voice recognition device, display controller, recorder device, display method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To notify of a voice-recognizable word to a user without spoiling the information transmission function using a display screen or the designability. <P>SOLUTION: A speech recognition part 110 can recognize a word to be recognized which is registered in a recognition dictionary 160. When a user makes a request for retrieval etc., a request processing part 120 generates constitution elements of a picture for displaying a processing result (retrieval result etc.) to the request. When there is a word to be registered in the recognition dictionary 160 among words included in such a screen, a display control part 130 determines a display style for the word, accordance with the display rules prepared for the words registered, beforehand in the recognition dictionary. Consequently, the words can be displayed, in a style different from words which are not registered in the dictionary; and while the result of processing that the user requests are shown, recognizable words can be indicated. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a display control device, a speech recognition device, a recorder device, a display method, and a program that display words that can be recognized by speech recognition.

  In recent years, voice recognition technology has been used in various devices, and voice input interfaces are incorporated in these devices. By using the voice input interface, it is possible to improve convenience when inputting to the device without using an input device such as a keyboard, but when the user utters a word or the like that is not subject to voice recognition, There is a possibility that a word different from the word intended by the user is erroneously recognized and an operation not intended by the user is performed.

  That is, in a general speech recognition technique, a word that is a target of speech recognition, that is, a recognizable word is stored in a recognition dictionary in advance, and a word that is closest to the input speech is stored among the words stored in the recognition dictionary. In order to output as a recognition result, if a word that is not stored in the recognition dictionary is uttered, the above-mentioned erroneous recognition may be made or may not be recognized.

  In particular, users who have rarely used a device equipped with such a voice input interface are often confused without knowing what the device is supposed to operate, and a manual for examining recognized words. However, the advantage of the voice input interface that simplifies input is not utilized.

  In order to solve the above problems, it is necessary to appropriately notify the user of the recognition target word using an output interface such as voice or display, and a list of recognizable words is displayed on a part of the display screen. And a technique for displaying a word that can be recognized by help utterance on an auxiliary screen has been proposed (for example, see Patent Document 2).

  In addition, when a word that can be input to the device is displayed in different display modes such as red and blue, and the type of display form of the word that the user wants to input, for example, “red”, “blue”, etc. A technique for inputting a word corresponding to the type of display form uttered has been proposed (see, for example, Patent Document 3).

JP-A-6-332665 JP-A-11-65739 Japanese Patent Laid-Open No. 2002-278587

  However, with the technique disclosed in Patent Document 1, it is difficult to display a list of all recognizable words at once if the number of words to be recognized becomes very large. The content that should be originally displayed cannot be displayed on the part. Furthermore, the design of the screen is also lost for displaying the list.

  In addition, in the technique disclosed in Patent Document 2, it is necessary to input a voice command such as a help utterance in order to display / erase the auxiliary screen. Therefore, redundant communication with the device increases, and input is simplified. The advantage of the voice input interface is not utilized.

  The technique disclosed in Patent Document 3 is a technique for inputting a desired word by uttering the type of display form. As a result, input items displayed can be input, but speech recognition is possible. The word itself is not notified to the user.

  That is, in the technique disclosed in Patent Document 3, the types of display forms such as “red” and “blue” are words that can be recognized by speech, but it is assumed that the user can input speech for these words. And an input item is assigned to a recognizable type of display form, and a word such as “red” that is known to be recognizable is uttered to select a corresponding input item. Therefore, words that can be recognized by voice cannot be notified to the user through the display, and since the input is not direct, there is a problem that it is difficult to understand because it is not intuitively linked with operation of a device equipped with a voice input interface. It was.

  The present invention has been made in view of the above, and a speech recognition device, a display control device, and the like that can notify a user of a speech-recognizable word without impairing an information transmission function and design by a display screen, An object of the present invention is to provide a recorder device, a voice recognition method, and a program.

In order to solve the above-described problems and achieve the object, a speech recognition apparatus according to one aspect of the present invention is input with a recognition dictionary that stores a plurality of words that are subject to speech recognition and how to read these words. Speech recognition means for performing speech recognition processing with reference to the recognition dictionary,
Display rule storage means for storing a display rule that defines a display mode of words stored in the recognition dictionary, and when the request processing result for the input request is displayed on the display screen, the recognition is displayed in the display screen. Display control means for determining a display mode according to a display rule stored in the display rule storage means for the included word when a word stored in the dictionary is included, To do.

  A speech recognition apparatus according to another aspect of the present invention refers to a recognition dictionary that stores a plurality of words to be speech-recognized and how to read these words, and refers to the recognition dictionary for input speech. A voice recognition unit that performs voice recognition processing, a display rule storage unit that stores a display rule that defines a display mode of words stored in the recognition dictionary, and a request based on the voice recognized by the voice recognition unit Included when a word stored in the recognition dictionary is included in the display screen when displaying a request processing result by the request processing means to the request processing means to be displayed on the display screen. Display control means for determining a display mode according to a display rule stored in the display rule storage means for the stored word.

  The display control apparatus according to another aspect of the present invention refers to a recognition dictionary that stores a plurality of words to be speech-recognized and how to read these words, and refers to the recognition dictionary for input speech. Display a word that can be recognized by a voice recognition device having voice recognition means for performing voice recognition processing, and when displaying a request processing result for an inputted request on a display screen, the display Display control means for controlling the contents, and display rule storage means for storing display rules for defining the display mode of the words stored in the recognition dictionary, the display control means should be displayed on the display screen When a word stored in the recognition dictionary is included in the result, the display mode is determined according to the display rule stored in the display rule storage unit for the included word. The features.

  A recorder apparatus according to another aspect of the present invention is a recorder apparatus that performs processing for storing an image, and includes a recognition dictionary that stores a plurality of words to be subjected to speech recognition and how to read these words, and an input Speech recognition means for performing speech recognition processing with reference to the recognition dictionary, request processing means for processing a request based on the speech recognized by the speech recognition means, and the request processing means for the request Display control means for controlling the display contents when the result is displayed on the display screen, and display rule storage means for storing a display rule for defining a display mode of words stored in the recognition dictionary, When the word to be stored in the recognition dictionary is included in the result to be displayed on the display screen, the control unit stores the included word in the display rule storage unit. And determining the display mode in accordance 憶 has been that display rule.

  In addition, a display method according to another aspect of the present invention includes a recognition dictionary that stores a plurality of words to be subjected to speech recognition and how to read these words, and refers to the recognition dictionary for input speech. A method for displaying words that can be recognized by a speech recognition device having speech recognition means for performing speech recognition processing, and displaying a request processing result for an input request on the display screen. If a word to be stored in the recognition dictionary is included in the result to be determined, a display mode is determined according to a display rule determined in advance for the included word.

The program according to another aspect of the present invention is used for speech recognition processing among the results to be displayed on the display screen when the computer displays the request processing result for the input request on the display screen. When a word to be stored is included in the recognition dictionary, the display dictionary is configured to function as a display control unit that determines a display mode according to a display rule determined in advance for the included word.

  According to the present invention, it is possible to notify a user of a word that can be recognized by voice without impairing the information transmission function and design of the display screen.

  Exemplary embodiments of a speech recognition device, a display control device, a recorder device, a speech recognition method, and a program according to the present invention will be explained below in detail with reference to the accompanying drawings.

(First embodiment)
FIG. 1 is a block diagram showing the configuration of the speech recognition apparatus according to the first embodiment of the present invention. As shown in the figure, the speech recognition apparatus 10 includes a speech input unit 100, a speech recognition unit 110, a request processing unit 120, a display control unit 130, a display unit 140, an acoustic dictionary 150, a recognition dictionary. 160, a rule application word table 170, a display rule storage unit 180, and a request processing information storage unit 190.

  As described above, the speech recognition apparatus 10 according to the present embodiment includes the request processing unit 120 and the request processing information storage unit 190 for inputting a request from the user and performing processing according to the request. The request processing unit 120 and the request processing information storage unit 190 do not need to be integrated into the voice recognition device, and may be separated from each other.

  The voice input unit 100 includes a microphone or the like, receives sound such as voice uttered by the user of the voice recognition device 10, and converts the received sound into an acoustic signal that can be processed by the voice recognition unit 110. Output.

  The speech recognition unit 110 analyzes the acoustic signal supplied from the speech input unit 100, refers to the acoustic dictionary 150, and requests a request processing unit as a recognition result that is acoustically most similar to a word stored in the recognition dictionary 160. 120 is output. Thus, when the user utters a word (request content or the like) to be input to the request processing unit 120 toward the voice input unit 100, the word to be input by the voice recognition unit 110 is input as a recognition result. Here, the word to be recognized includes not only one word but also a word string composed of a plurality of words. As the voice recognition method performed by the voice recognition unit 110, various known methods such as a method using an HMM (Hidden Markov Model) can be used.

  The acoustic dictionary 150 stores information related to the sound used by the speech recognition unit 110 as described above, and the recognition dictionary 160 recognizes words that can be recognized by the speech recognition unit 110 and recognizable words. A reading that can be uttered by the user is stored. FIG. 2 shows an example of information stored in the recognition dictionary 160.

  As shown in the figure, the recognition dictionary 160 includes the recognition target words “Information Communication”, “Shin”, “Huangmen-sama”, “High School Baseball”, etc., and the reading “Jouhou” corresponding to these recognition target words. “Tsu”, “Shin”, “Koumon-sama”, “Koukou-yakyu”, etc. are stored in correspondence. Therefore, when the user utters “Komon-sama” to the voice input unit 100, the speech recognition unit 110 recognizes the word “Komon-sama” corresponding to the reading “Koumon-sama” and outputs it as a recognition result. is there.

  Returning to FIG. 1, the request processing unit 120 performs processing according to the user request, which is the recognition result of the voice recognition unit 110 as described above, and displays the processing result to display the processing result on the display unit 140. Output to the control unit 130. In the present embodiment, the request processing unit 120 performs a search process on a word (such as a TV program name) input by the user, and the request processing unit 120 responds to the request in the request processing information storage unit 190. Information necessary for processing is stored.

  An example of information stored in the request processing information storage unit 190 is shown in FIG. In the example shown in the figure, the request processing information storage unit 190 stores a program database necessary for the search processing performed by the request processing unit 120. When such a program database is stored, the request processing unit 120 can search for a television program in response to a program search request from the user.

  The program database is associated with item information relating to programs such as “ID”, “program name”, “broadcast date”, “broadcast station”, “genre”, and “performer”. By referring to such a program database, when the user makes a search request using a performer name (such as Ryutaro Mine) or a genre (society / report) as a key, the request processing unit 120 appears. It is possible to search for a program that is currently being executed or a program of a requested genre.

  That is, in the present embodiment, a word such as a program name or a performer name for which a user requests a search is issued to the voice input unit 100, so that the word is recognized and supplied to the request processing unit 120. . The request processing unit 120 searches the information stored in the request processing information storage unit 190 for information related to words input using speech recognition in this way, and displays a search result. Elements are supplied to the display control unit 130.

  Returning to FIG. 1, the display control unit 130 generates display data for displaying on the display unit 140 the result of the user request processed by the request processing unit 120, that is, the search result related to the word requested by the user. And output to the display unit 140. The display control unit 130 according to the present embodiment does not simply display the processing result of the request processing unit 120, but can display the word stored in the recognition dictionary 160, that is, can recognize the voice when displaying the processing result. The display content is controlled so that a display that can notify the user of the word is performed.

  In order to perform the display content control as described above, the display control unit 130 includes a rule application word determination unit 131 and a display mode determination unit 132. The rule application word determination unit 131 stores the words stored in the recognition dictionary 160 among the processing results (such as the search result of the TV program) supplied from the request processing unit 120 to be displayed on the display unit 140 as described above. (“Information communication”, “Huangmen”, etc. in FIG. 2) is investigated, and if it is included, the rule is applied because the word is applied to the display rule described later. Register in the word table 170.

  Here, an example of the contents of the rule application word table 170 is shown in FIG. As shown in the figure, in the rule application word table 170, a word included in the processing result by the request processing unit 120 and an application flag indicating whether or not the display rule is applied to the word are registered. . Here, when the application flag is “1”, it indicates that there is application, and “0” is a flag indicating that there is no application. As described above, the application flag “1” is used for the words “Huangmen” and “next page” stored in the recognition dictionary 160, and the application flag “2” is not stored in the recognition dictionary. “0” is given.

  The display mode determination unit 132 refers to the rule application word table 170 registered by the rule application word determination unit 131 as described above, and the application flag is “1”, that is, for the word that is determined to apply the rule, The display content is controlled so that display is performed in a manner according to the display rules stored in the display rule storage unit 180. In other words, special processing is not performed for words to which the rule is not applied, and is displayed in a mode according to the display setting of the device, whereas for words registered that the rule is applied, Regardless of the display setting, the display rule is controlled to be displayed in a manner according to the display rule stored in the display rule storage unit 180.

  Here, FIG. 5 shows an example of display rules stored in the display rule storage unit 180. As shown in the figure, in this display rule, in the case of a word to which the application flag “0” is assigned, that is, a word that is not applied is displayed as “no change”, that is, in a mode according to the display setting of the device. On the other hand, in the case of a word to which the application flag “1” is assigned, that is, for an applied word, it is defined that it should be displayed in a form such as “Font + 2, Bold”. Since the display rule in the claims refers to a rule applied to words stored in the recognition dictionary 160, the rule corresponding to the application flag “1” stored in the display rule storage unit 180 is the claim. It corresponds to the display rule in the range. However, another rule to be applied to the word corresponding to the application flag “0” may be prepared and stored in the display rule storage unit 180.

  The display mode determination unit 132 that refers to this displays the word with the application flag “1” two times larger than the normal setting and makes it bold, that is, the font is larger than the word that is not applied. Is controlled to be displayed two times larger and bold.

  The display unit 140 has a display screen such as an LCD (Liquid Crystal Display), and the content corresponding to the processing result of the request processing unit 120 on this display screen is the content controlled by the display control unit 130. Display is made.

  The above is the configuration of the speech recognition apparatus 10 according to the first exemplary embodiment of the present invention. Hereinafter, the operation of the speech recognition apparatus 10 will be described with specific examples. Here, the request processing unit 120 performs a process for a TV program search request by voice input from the user, for example, a program search request specifying a genre or a performer, and displays the TV program search result as the processing result on the display unit 140. An example of the case will be described.

  First, the user utters information for a desired search toward the voice input unit 100, that is, information such as the genre of the desired program, performers, channels, broadcast time, program name, and the like. For example, a search condition such as “today's drama”, “10 channels”, “a program in which Takeshi Matsudaira appears” is uttered. Then, the voice input unit 100 converts the voice into an acoustic signal, and the voice recognition unit 110 performs voice recognition processing on the acoustic signal.

  Here, when the user utters a word such as “today's drama”, “today's drama” is recognized by the voice recognition unit 110 and is supplied to the request processing unit 120 as a search condition wording. The request processing unit 120 performs a search process according to the search condition wording. A procedure of such search processing will be described with reference to FIG.

  As shown in the figure, the request processing unit 120 determines whether or not the input “today's drama” is acceptable, and at the same time, selects a word corresponding to a search key or an operation command and its attribute in the program search process. Extract (step S101). For example, it is assumed that the request processing unit 120 holds an acceptable word string in the form of a template in advance, and determines whether or not it is acceptable depending on whether or not it matches the template. Here, it is assumed that “genre” of “date” of a plurality of templates held by the request processing unit 120 is included.

  In this case, it can be determined that “Today” of the recognition result is a date and “Drama” is a specific value of the genre, and “Drama” of “Today” matches the template held. Therefore, “Today's Drama” is determined as an acceptable word, and the attribute “Today” and the attribute “Drama” are extracted.

  When the attribute is extracted as described above, the request processing unit 120 creates a search query using the extracted attribute value, and creates a condition from the program database (see FIG. 3) stored in the request processing information storage unit 190. The program that matches is searched (step S102). For example, if the date of the request is August 20, 2003, a program such as “shin” with ID “02” and “Komon-sama” with ID “03” is searched.

  When the program search is performed as described above, the request processing unit 120 generates content for displaying the search result on the display unit 140, that is, a component of the display screen (step S103). Here, the request processing unit 120 generates a display screen component using a template of the display screen that is held in advance. Here, FIG. 7 shows an example of a held template.

  The request processing unit 120 generates the constituent elements of the display screen in such a manner that the search result or the like is applied to the frame represented by [] of the template as shown in FIG. Then, when the programs “shin” and “yellow gate-like” are retrieved as described above, the components of the display screen as shown in FIG. 8 are generated.

  The above is the search processing by the request processing unit 120 and the generation processing of the component for displaying the result. The components of the display screen of the search result generated in this way are supplied to the display control unit 130. The display control unit 130 does not display the word included in the component for displaying the search result as described above, but informs the user of a word that can be recognized when the search result is displayed. The display content is controlled so that a clear display is performed.

  The processing operation of the rule application word determination unit 131 of the display control unit 130 that performs such control will be described by taking as an example the case where components of the search result display screen as shown in FIG. 8 are supplied. As illustrated in FIG. 9, the rule application word determination unit 131 acquires the first word (“2 cases”) from the components of the search result display screen (step S <b> 201).

  As a method of cutting out words from the constituent elements of the display screen in this manner, various known methods such as a method of cutting out a character string delimited by white space as one word and a morphological element analysis method can be used. When the first word is cut out in this way, the acquired word (“2 cases”) is registered in the rule application word table 170, and the application flag corresponding to the word is initialized (flag = “0”) (step). S202).

  Then, the first word (“information communication”) stored in the recognition dictionary 160 (see FIG. 2) is acquired (step S203). Thereafter, the word acquired from the components of the display screen, that is, the word acquired from the words included in the content to be displayed as the search result is compared with the word acquired from the recognition dictionary 160, and whether or not they match. It is determined whether or not (step S204).

  Here, if the two do not match (in the above example, the acquired words are “2” and “informative”, they do not match), the words acquired from the constituent elements of the display screen are stored in the recognition dictionary 160. It is determined whether or not all the words that have been compared have been compared (step S205). If not all words are compared, the next word is acquired from the recognition dictionary 160 (step S206), and the process returns to step S204 to return the word acquired from the recognition dictionary 160 and the components of the display screen. It is determined whether or not the word acquired from the above matches. That is, it is determined whether or not the word included in the content to be displayed as the search result is a word stored in the recognition dictionary 160.

  On the other hand, if it is determined in step S204 that both words match, “1” is set in the application flag of the word in the rule application word table 170 (step S207), and registration that the word is a rule application word is performed. . Then, it is determined whether or not all the words included in the constituent elements of the display screen have been compared with the words stored in the recognition dictionary 160 (step S208). If all the words have not been processed, the display screen The next word is acquired from the component (step S209), the process returns to step S202, and is registered in the rule application word table 170, and the application flag corresponding to the word is initialized.

  If it is determined in step S205 that the comparison of all the words included in the components of the display screen with all the words in the recognition dictionary 160 has been completed, the process of setting the application flag (step S207) is not performed. Proceed to S208. That is, the application flag remains “0” for the certain word in the constituent elements of the display screen.

  If it is determined in step S208 that the comparison processing has been completed for all the words included in the constituent elements of the display screen, the registration processing in the rule application word table 170 is terminated. That is, when it is determined whether all the words included in the constituent elements of the search result display screen match all the words in the recognition dictionary 160, the process ends.

  By performing the processing as described above, it is determined whether or not the rule should be applied to the contents to be displayed of the search result of the request processing unit 120, that is, all the words included in the constituent elements of the display screen. For a word to which a rule is applied, a flag “1” to that effect is registered in the rule application word table 170.

  When the registration process to the rule application word table 170 as described above is completed, the display mode determination unit 132 refers to the rule application word table 170 after the registration process, and the configuration of the display screen supplied from the request processing unit 120 Processing for determining the display mode of each word in the element is performed. Details of this processing will be described with reference to FIG.

  As shown in the figure, first, the first word registered in the rule application word table 170 is acquired (step S301). Then, referring to the display rule (see FIG. 5) stored in the display rule storage unit 180, a rule corresponding to the application flag corresponding to the acquired word is extracted (step S302). That is, when the application flag is “1”, the display rule for the word corresponding to the application flag “1”, that is, the word registered in the recognition dictionary 160 is extracted for the word, and the display correction according to the rule is performed. I do. Specifically, the display form is corrected so that the font is enlarged by two and bolded (step S303).

  On the other hand, when the application flag corresponding to the acquired word is “0”, it means that the extracted rule is “no change”, that is, no special display rule is applied, and there is no special correction according to the rule. Will not be. In this embodiment, the information is stored as no change, but the application flag “0” means that the display rule is not applied. Therefore, such information may not be stored.

  When correction according to the display rule is performed as described above (there may be no correction), whether or not correction processing according to the display rule has been performed for all the words registered in the rule application word table 170. Is determined (step S304). If the processing has not been completed for all the registered words, the next word is acquired from the rule application word table 170 (step S305), and the processing after step S302 is performed.

  On the other hand, when the process is completed for all registered words, the display mode determination process is terminated. By performing such processing, the contents to be displayed for the search result of the request processing unit 120, that is, the words stored in the recognition dictionary 160 among the words included in the constituent elements of the display screen, are displayed in the display rule. Accordingly, a display mode correction process (a font 2 size larger than the case where no rule is applied and bold) is performed.

  As described above, the display content is displayed by the display control unit 130 such that the words that can be recognized by voice among the words included in the constituent elements of the search result display screen are displayed in a mode (font, font) different from other words. Be controlled. Then, a display screen of contents controlled by the display control unit 130 is displayed on the display unit 140. Here, an example of the display content of the display unit 140 is shown in FIG. As shown in the figure, among words on the TV program search result display screen, words stored in the recognition dictionary 160, that is, words that can be recognized by speech (such as broadcast station names such as “NBS” and “SBS”, “ Genre names such as “Drama” and words such as program names such as “Shin” and “Huangmen” are bolded and highlighted in a font that is two sizes larger.

  As described above, in the present embodiment, when the user makes a processing request, the request processing unit 120 processes the request, and the processing result (in the above example, the search result of the TV program) is displayed on the display unit 140. Is done. And the word which can be recognized with the said speech recognition apparatus 10 among the words contained in such a display screen can be displayed in the aspect different from another word (refer FIG. 11). As a result, the user who refers to the display can know the result of his request and can know a word that can be recognized by voice.

  Therefore, it is not necessary to display a special list to inform the user of words that can be recognized by voice, or to display an auxiliary screen with help utterances, and it is possible to recognize voice without disturbing provision of necessary information. Simple words can be notified to the user. In addition, since a special screen or the like for notifying a recognizable word is not required, it is possible to prevent the screen design from being greatly impaired.

  In particular, users who are unfamiliar with voice recognition or users who have hardly used a voice recognition-equipped device itself are confused because it is difficult to understand what the target device will work by looking at the screen. There are many cases. In the present embodiment, recognizable words are displayed (highlighted) in a different manner from others and presented to the user. In addition, when any display screen is displayed, among the words included in the display screen as described above, words that can be recognized by voice are highlighted and fixed, that is, while the speech recognition device is being used. Since recognizable words are consistently highlighted, useless work such as re-reading the manual to check the recognized words is reduced. Furthermore, a great effect can be expected, such as a high possibility of reducing malfunctions caused by the input of unknown words.

(Second Embodiment)
Next, a second embodiment of the present invention will be described. FIG. 12 is a block diagram showing a configuration of the speech recognition apparatus 20 according to the second exemplary embodiment of the present invention. As shown in the figure, in addition to the configuration of the speech recognition device 10 in the first embodiment, the speech recognition device 20 in the second embodiment includes a reading frequency management unit 210, a display reading determination unit 220, and the like. And a reading history storage unit 230, and a recognition dictionary 260 in place of the recognition dictionary 160 and a rule application word table 270 in place of the rule application word table 170, differing from the first embodiment. Yes. In the second embodiment, the same reference numerals are given to the same components as those in the first embodiment, and the description thereof is omitted.

  In the recognition dictionary 260 in the present embodiment, a plurality of readings are associated with one recognition target word. Here, FIG. 13 shows an example of data stored in the recognition dictionary 260. As shown in the figure, the recognition dictionary 260 has three words “Montsu Hysteria Theater“ Indictment Defense Lawyer Series: Fumiaki Inokuma 5 ”” in three ways to read “Getsuyo Histeri Gejo” and “Koku”. "Hatsubengo Ninshirizu" and "Inokuma Fumiaki Faibu" are associated. The speech recognition unit 110 that refers to the recognition dictionary 260 recognizes the recognition target word “Monday hysterical theater“ accusal defense attorney series, Inokuma Bunmei 5 ”” when the user speaks in any of the above three readings. can do.

  A result recognized by the voice recognition unit 110 is supplied to the reading frequency management unit 210. The reading frequency management unit 210 adds, to the reading history storage unit 230, a combination of a word (a recognition target word in FIG. 2) included in the recognition result of the voice recognition unit 110 and a reading uttered by the user at the time of recognition. Update as you do. That is, in this embodiment, as described above, a plurality of readings are associated with one recognition target word, and the user can recognize the recognition target word by managing the combination appearance frequency as described above. Can understand how to read the most.

  The reading history storage unit 230 stores the frequency of occurrence of combinations of recognition target words and readings managed by the reading frequency management unit 210 as described above. Here, an example of the storage contents of the reading history storage unit 230 is shown in FIG. As shown in the drawing, the reading history storage unit 230 stores information such as a reading-ready character string in association with the combination of the recognition target word and the reading and the frequency thereof. Here, the reading-compatible character string is a character string in a section corresponding to the reading.

  When speech recognition is performed by the speech recognition unit 110, the frequency of the reading history storage unit 230 as described above is updated by the reading frequency management unit 210 according to the recognition content. For example, if the user utters “Nihonmarue” in the state where the memory is as shown in the figure, and the recognition target word “Nihon Maru! TV correspondent” is recognized according to the utterance, The frequency corresponding to “Marumie” is updated from “5” to “6”.

  The display / reading determination unit 220 refers to the content stored in the reading history storage unit 230, and selects a rule application target character string for a word having a plurality of readings among the words registered in the rule application word table 270. The writing process is performed according to a predetermined standard. The recognition target words and how to read the rule application word table 270 are registered by the rule application word determination unit 131 in the same procedure as in the first embodiment.

  Here, an example of the registered contents of the rule application word table 270 in the present embodiment is shown in FIG. As shown in the figure, the rule application word table 270 in the second embodiment is obtained by adding an item of a rule target character string in addition to the table contents of the first embodiment (see FIG. 4). The display / reading determination unit 220 performs a process of writing a rule target character string determined according to a predetermined criterion in the item.

  In the present embodiment, a word having a plurality of readings registered in the rule application word table 270 (for example, “Nippon Maruhi! TV correspondent”) is referred to the reading history storage unit 230 and read most frequently ( The character string corresponding to “Nihon Marumi”) (“Nippon Maru Appearance”) is written into the rule target character string.

  The display mode determination unit 132 in the second embodiment refers to the rule application word table 270 registered after the rewritten word as described above, and similarly to the first embodiment, the display rule storage unit The display mode is determined based on the display rule stored in 180.

  Note that the request processing information storage unit 190 in the second embodiment stores a program database for searching for television programs, as in the first embodiment, and an example of the contents thereof is shown in FIG. 16 shows. As shown in the figure, this program database includes information on items related to TV programs such as ID, program name, broadcast date and time, broadcast station, genre, and performers, as in the first embodiment. As described above, a recognition target word having a plurality of readings (such as “Monday hysterical theater“ accusal defense counsel series / Fumiaki Inokuma 5 ”) is also included.

  The above is the configuration of the speech recognition apparatus 20 according to the second exemplary embodiment. Hereinafter, the operation of the speech recognition apparatus 20 will be described with specific examples. Here, the request processing unit 120 performs a process for a TV program search request by voice input from the user, for example, a program search request specifying a genre or a performer, and displays the TV program search result as the processing result on the display unit 140. An example of the case will be described.

  First, the user utters information for a desired search toward the voice input unit 100, that is, information such as the genre of the desired program, performers, channels, broadcast time, program name, and the like. Here, it is assumed that the word “Getsuyo Histeri Gekijo” is uttered in order to recognize the recognition target word “Monday Hystery Theater Prosecution Attorney Series, Fumiaki Inokuma 5” having a plurality of readings.

  In this case, the “Monday Hystery Theater Prosecution Attorney Series / Fumiaki Inokuma 5” is recognized by the voice recognition unit 110 and supplied to the request processing unit 120 as a search condition wording. The request processing unit 120 performs a search process according to the search condition wording.

  When the voice recognition unit 110 performs the voice recognition as described above, the reading frequency management unit 210 updates the storage content of the reading history storage unit 230 according to the recognition result. The processing of the reading frequency management unit 210 when the speech recognition unit 110 performs speech recognition in this way will be described with reference to FIG.

  First, the recognition result of the first word by the speech recognition unit 110, that is, the recognition target word and how to read it at the time of the recognition are acquired (step S401). Here, the recognition target words “Monday Hystery Theater Prosecution Attorney Series, Fumiaki Inokuma 5” and the reading “Getsuyo Histeri Gekijo” are acquired.

  Next, how many times the combination frequency of the recognition target word and the reading that is the acquired recognition result, that is, the combination frequency of “Monday Hystery Theater Prosecution Defense Attorney Series ・ Fumaaki Inokuma 5” and “Getsuyo Histeri Gekijo” Information indicating whether it has appeared is acquired from the history storage unit 230 (see FIG. 14) (step S402).

  Then, 1 is added to the acquired frequency, the value after the addition is overwritten on the frequency corresponding to the combination, and the storage content of the reading history storage unit 230 is updated (step S403). In the state as shown in FIG. 14, the frequency “1” corresponding to the combination of the recognition target word and the reading “Getsuyo Histeri Gekijo” is acquired, and 1 is added to this. The frequency is updated to “2”.

  When the update is performed in this way, it is determined whether or not the above-described processing has been performed on all the words included in the recognition result of the speech recognition unit 110 (step S404). The process ends. On the other hand, if not all the words have been processed, the next recognition target word and reading combination included in the recognition result of the speech recognition unit 110 is acquired (step S405), and the processing after step S402 is performed for the combination. I do. In the case of the above example, since the user only utters the word “Getsuyo Histari Gekijo”, there is only one recognition target word, the determination in Step S404 is “Yes”, and the reading frequency management unit The processing of 210 ends.

  Further, the recognition result by the voice recognition unit 110 as described above is supplied to the request processing unit 120, and the component for displaying the processing result according to the request included in the recognition result, as in the first embodiment. Is output to the display control unit 130. Receiving this, the rule application word determination unit 131 registers the words included in the constituent elements of the display screen in the rule application word table 270 together with the application flag, as in the first embodiment.

  In the second embodiment, unlike the first embodiment, the display / reading determination unit 220 performs a process of writing the rule target character string for the words registered in the rule application word table 270 as described above. Processing performed by the display reading determination unit 220 will be described with reference to FIG. Here, the request processing unit 120 creates a component for displaying the processing result as shown in FIG. 19, and as a result, the rule application word table 270 is registered as shown in FIG. I will explain.

  First, the display reading determination unit 220 acquires “two cases” that are the first words stored in the rule application word table 270 (step S501). Then, all the words (“2 cases”) are provisionally registered as they are in the item of the rule application target character string corresponding to the acquired word in the rule application word table 270 (step S502).

  Then, the first recognition target word (in the example shown in FIG. 14, “Nippon Maru looks! TV correspondent”) stored in the reading history storage unit 230 is acquired (step S503) and acquired from the rule application word table 270. It is compared with the word (“2 cases”) and it is determined whether or not they match (step S504).

  If the two do not match, such as “two cases” and “Japan Marutsu! TV correspondent”, it is determined whether or not all the words stored in the reading history storage unit 230 have been compared (steps). S505) If the comparison with all the words has not been completed, the next word stored in the reading history storage unit 230 is acquired (step S506), and the processes after step S504 are performed.

  On the other hand, if both words match in the determination of step S504, refer to a plurality of readings for the word stored in the reading history storage unit 230 and the frequencies corresponding to the respective readings (and reading corresponding character strings), A reading correspondence character string corresponding to the reading with the highest frequency value for the word is acquired and stored in the rule application target character string of the rule application word table 270 (step S507). In the above example, the reading-ready character string “Nippon Maru Appearance” corresponding to the reading “Nihon Maru Mie” with the highest frequency value (5 times) is obtained from the multiple readings of the word “Japan Maru Appearance! TV correspondent”. Then, it is stored in the rule application target character string corresponding to the word “Nippon Maru Appearance! Television Correspondent” in the rule application word table 270.

  As described above, when a character string corresponding to a high-frequency reading is stored in the rule target character string, or when all the words stored in the reading history storage unit 230 have been compared in step S505, the rule It is determined whether or not the processing such as storing the rule application target character string as described above has been performed on the words registered in the application word table 270 (step S508). If all the words have been processed, the processing is performed. Exit. On the other hand, if all the words registered in the rule application word table 270 have not been processed, the next word is acquired from the rule application word table 270 (step S509), and the processes after step S502 are performed.

  In this way, processing such as writing a character string into the rule application target character string is performed for all words registered in the rule application word table 270. An example of the registered contents of the rule application word table 270 after processing by the display reading determination unit 220 is shown in FIG. As shown in the figure, for the word “2 cases” and the like that are not registered in the recognition dictionary 260, the rule application target character string remains as “2 cases” temporarily registered, but there are a plurality of reading methods. For the word “Nippon Maru Appearance! TV correspondent” registered in the recognition dictionary 160, the character string “Japan Maru Appearance” corresponding to the most frequent reading of the recognizable readings is registered as the rule application target character string. The Note that “super TV / information front row” is also a word having a plurality of readings, but in the illustrated example, the most frequent reading is a reading corresponding to the whole word “Super Television”. The character string corresponding to the reading, that is, all the words are registered.

  As described above, when the word, the rule application character string, and the application flag are registered in the rule application word table 270, the display mode determination unit 132 refers to the rule application word table 270 after the registration process, and the request processing unit 120 The process which determines the display mode of each word in the component of the display screen supplied from is performed. Details of such processing will be described with reference to FIG.

  In this description, a case where rules such as those shown in FIG. 22 are stored in the display rule storage unit 180 will be described as an example. That is, in this display rule storage unit 180, in addition to the display rule applied to the character string or the like whose application flag is “1”, the display rule to be applied to the character string or the like whose application flag is “0”. Is stored, and the display mode determination unit 132 determines the display mode according to the rules stored in the display rule storage unit 180.

  The processing performed by the display mode determination unit 132 is basically the same as that in the first embodiment (see FIG. 10) (steps S601 to S602 correspond to steps S301 to S302, and steps S604 to S604). Step S605 corresponds to Step S304 to Step S305), but the processing of Step S603 is different from that of the first embodiment.

  That is, it is the same until the word registered in the rule application word table 270 is acquired (steps S601 and S605) and the display rule corresponding to the application flag of the word is extracted (step S602). The character strings that determine the display mode using the display rules are different. More specifically, in the first embodiment, the display mode is determined according to the display rule for all parts of the word, but in this embodiment, the display mode is stored in the rule application word table 270. The display mode is determined and corrected according to the display rule for the rule application target character string (step S603).

  For example, as shown in FIG. 23, if the word extracted from the rule application word table 270 is “Japan Maru looks! Television correspondent”, the application flag is “1” and the rule application target character string is “Japan Maru Appearance”. Therefore, according to the display rule corresponding to the application flag “1”, that is, the display rule applied to the word registered in the recognition dictionary 260, the font size of only the part “Japan looks!” Is increased by two. And bolded. On the other hand, in the case of “2 cases” taken out, the application flag is “0”, and the rule application target character string is “2 cases” that is all the words, so the display rules for the words registered in the recognition dictionary 260 are displayed. Does not apply, and rules that apply to other words apply. Therefore, the font size is displayed two smaller.

  In addition, when the acquired word is “super TV / information front row”, the application flag is “1”, so the display rule for the word registered in the recognition dictionary 260 is applied. Since the column is the entire word portion “super TV / information front row”, the result is that the font size of the entire word is two larger and bolded as in the first embodiment.

  As described above, in the present embodiment, when the user makes a processing request, the request processing unit 120 processes the request, and the processing result (in the above example, the search result of the TV program) is displayed on the display unit 140. Is done. And among the words contained in such a display screen, the character string necessary for recognizing the word that can be recognized by the voice recognition device 20 is different from other characters. It can be displayed (see FIG. 23). As a result, the user who has referred to the display can know the result of his / her request, can know a word that can be recognized by voice, and can recognize the word by how to read it. You can know if you can.

  Therefore, as in the first embodiment, it is possible to notify a user of a recognizable word while preventing the provision of necessary information and suppressing the screen design from being greatly impaired. Further, it is possible to notify the reading necessary for recognizing the word.

  Furthermore, in this embodiment, as a way of reading a word necessary for recognizing a recognizable word, a character string corresponding to the reading most frequently used by the user is displayed in a different form from the other, so that it is most used. Users can be told how to read easily. That is, when there is a difference in reading depending on the user, by emphasizing not only the whole word but only a necessary part, recognizable words can be distinguished from other words not recognized and displayed more clearly. Moreover, when a word itself is long like a program name, it is troublesome to utter the whole word, and it is particularly troublesome if a long word has to be uttered many times. Therefore, as in this embodiment, it is necessary to recognize the entire word by uttering the partial character string and to recognize the long word many times by notifying the partial reading. In this case, it is particularly suitable.

(Third embodiment)
Next, a third embodiment of the present invention will be described. FIG. 24 is a block diagram showing a configuration of a speech recognition apparatus 30 according to the third exemplary embodiment of the present invention. As shown in the figure, in addition to the configuration of the speech recognition device 10 in the first embodiment, a speech recognition device 30 in the third embodiment includes a word importance degree determination unit 310 and a word importance degree storage unit. 320 and an importance determination rule storage unit 330, and a display rule storage unit 380 instead of the display rule storage unit 180 in the first embodiment. Note that, in the third embodiment, components that are the same as those in the first embodiment are assigned the same reference numerals, and descriptions thereof are omitted.

  The word importance determination unit 310 is supplied with the processing result of the request processing unit 120 corresponding to the request from the user. The word importance level determination unit 310 determines the importance level for the word included in the component of the processing result display screen according to the importance level determination rule stored in the importance level determination rule storage unit 330, and uses this as the word importance level. It is stored in the degree storage unit 320. The word importance level determination unit 310 refers to the information stored in the request processing information storage unit 190, the processing content performed in the past by the request processing unit 120, etc., if necessary for such importance level determination. To do.

  The importance determination rule storage unit 330 stores rules for determining the importance of words as described above. Here, FIG. 25 shows an example of rules stored in the importance determination rule storage unit 330. As shown in the figure, the importance determination rule includes an importance value and the corresponding rule content.

  The illustrated example is an example of a rule when the request processing unit 120 performs a process of searching for a TV program using a genre or the like as a keyword. If the attribute of a word for which importance is determined is an already input attribute, the example shown in FIG. For example, a rule that the importance is determined to be 20 or a rule that the importance is determined to be 80 if the attribute of the target word is a sub-genre attribute when the genre attribute is input immediately before. .

  For example, if the word whose importance is to be determined is the search keyword “sports” of a television program, and the same word “sports” has been input to the request processing unit 120 before that, the same is repeated again. Since it is considered that there are few cases where a word is input, the importance of such a word is determined to be low. On the other hand, when a genre name “sports” or the like is input as a search keyword immediately before, a sub-genre such as “baseball” or “soccer” that is a subordinate genre is highly likely to be input as a search keyword. Therefore, when the word has such an attribute, the importance of the word is determined to be high.

  As described above, the importance determination rule also takes into account past input contents and the like, and the words categorized in advance according to the possibility of input from the user when the request processing unit 120 performs processing. The contents of the rules such as attributes and the importance corresponding to them are associated with each other. By determining the importance of the words using such rules, it is possible to It is possible to determine an importance level such that a word having a high possibility of being input has a high importance and a word having a low possibility of being input has a low importance.

  The importance determined according to such importance rules is stored in the word importance storage unit 320. Here, FIG. 26 shows an example of the stored contents of the word importance storage unit 320. As shown in the figure, in the word importance storage unit 320, the importance of each word determined by the word importance determination unit 310 by referring to the importance rules as described above is associated with each word. Stored.

  The display mode determination unit 132 in the present embodiment stores the words registered in the rule application word table 170 by the rule application word determination unit 131 in the display rule storage unit 380, as in the first embodiment. The display mode is determined based on the display rules and the word importance stored in the word importance storage unit 320.

  Here, an example of the contents stored in the display rule storage unit 380 in the present embodiment is shown in FIG. As shown in the figure, the display rule storage unit 380 stores four rules. First, the first rule is a rule used for a word (application flag “0”) that is not registered in the recognition dictionary 160. For the word, it is defined that the font is reduced by two.

  The other three rules are display rules that are applied to words registered in the recognition dictionary 160 (application flag “1”), and three types of rules are defined according to the importance range. When the importance is 50 or more and less than 80, the font is increased by two and displayed in bold. When the importance is 80 or more, the font is increased by four and displayed in bold. It is stipulated. Further, when the importance is less than 50, “None”, that is, not changing to the normal state is defined. As described above, in this embodiment, a rule is set such that the higher the importance, the more emphasized display is made.

  In accordance with the rules stored in the display rule storage unit 380, the display mode determination unit 132 in the present embodiment displays the display mode for words included in the components of the screen that displays the processing result of the request processing unit 120. Make corrections. In this way, an image for displaying the processing result of the request processing unit 120 including the word registered in the recognition dictionary 160 and the like whose display mode is corrected according to the importance is displayed on the display unit 140.

  The above is the configuration of the speech recognition apparatus 30 according to the third exemplary embodiment. Hereinafter, the operation of the speech recognition apparatus 30 will be described with specific examples. Here, the request processing unit 120 performs a process for a TV program search request by voice input from the user, for example, a program search request specifying a genre or a performer, and displays the TV program search result as the processing result on the display unit 140. An example of the case will be described.

  First, the user utters information for a desired search toward the voice input unit 100, that is, information such as the genre of the desired program, performers, channels, broadcast time, program name, and the like. Here, it is assumed that the utterance “Kyono Sports” is used to recognize the genre “Today's Sports” as a search keyword.

  In this case, “today's sport” is recognized by the voice recognition unit 110 and supplied to the request processing unit 120 as a search keyword. The request processing unit 120 performs a search process according to the search keyword, and as a result, a component of a screen on which a process result as shown in FIG. 28 is to be displayed is created.

  It should be noted that the request processing unit 120 in the present embodiment includes a plurality of genres of higher concepts, such as “sports”, and subgenres of lower concepts belonging to the genre of higher concepts, such as a subgenre “baseball” of the genre “sports”. It is possible to search the genre of the hierarchy as a search keyword. Therefore, as shown in FIG. 29, the request processing unit 120 according to the present embodiment holds a template of display screen components different from the template according to the first embodiment (see FIG. 7). Generate a component of the display screen using a template.

  As shown in the figure, in addition to the items of the template of the first embodiment, such a template is provided with a region where “sub-genre” is to be arranged. Accordingly, as shown in FIG. 28, the “genre” item of the template is applied to “sports” that is a genre of the higher concept, and “subgenres” are “baseball” and “soccer” that are genres of the lower concept. "Is applied, the display screen is configured.

  The processing result by the request processing unit 120 is also supplied to the word importance determining unit 310, and the processing contents of the past from the request processing unit 120 (the values of the genre attribute and date attribute have already been determined as a program search query) Information) is provided. The word importance level determination unit 310 performs a process of determining the importance level for a word included in the processing result, and the contents of the processing will be described with reference to FIG.

  First, the first word is acquired from the words included in the components of the processing result display screen created by the request processing unit 120 and registered in the recognition dictionary 160 (step S701). For example, if the contents as shown in FIG. 31 are registered in the recognition dictionary 160 and the display screen is as shown in FIG. 28, “J League Kashima × Tokyo” is acquired.

  Next, the acquired word is registered in the word importance storage unit 320, and the importance corresponding to the acquired word is initialized (step S702). Thereafter, the first rule content is acquired from a plurality of rule contents (see FIG. 25) stored in the importance determination rule storage unit 330 (step S703), and whether the acquired word satisfies the acquired rule content It is determined whether or not (step S704).

  Here, if the acquired word does not satisfy the acquired rule contents, it is determined whether or not all the rule contents stored in the importance determination rule storage unit 330 have been determined (step S705), and all the rules are determined. If the content is not judged, the next rule content stored in the importance determination rule storage unit 330 is acquired (step S706), and it is determined whether or not the rule content is satisfied (step S704).

  On the other hand, when the acquired word satisfies the acquired rule content, the importance level of the word subject to importance determination is determined as the importance value corresponding to the rule content, and the word importance level storage is performed. The data is stored in the unit 320 (step S707). For example, if the target word has already been input to the request processing unit 120, that is, if the same word has already been input, the rule content corresponding to the importance 20 is satisfied. 20 ”is stored in the word importance storage unit 320. When “J League Kashima × Tokyo” is the target word, the attribute of the word is a program name, and is not a genre attribute or a date attribute, so it is determined that the rule content is not satisfied.

  When the importance of the word is determined and stored in the word importance storage unit 320 as described above, or when it is determined in step S705 that the determination as to whether or not all the rule contents are satisfied is completed. It is determined whether or not the processing such as importance determination has been completed for all the words registered in the recognition dictionary 160 and included in the constituent elements of the display screen (step S708).

  Here, when the processing is completed for all the words, the processing by the word importance determining unit 310 ends. On the other hand, if the processing has not been completed for all the words, unprocessed words that are words included in the constituent elements of the display screen and registered in the recognition dictionary 160 are acquired (step S709). Then, the processing after step S702 is performed on the word. In this way, importance determination processing is performed for all the words included in the components of the display screen and registered in the recognition dictionary 160, and the result is stored in the word importance storage unit 320.

  As described above, the degree of importance of the word is determined and stored in the word importance degree storage unit 320. Further, as in the first embodiment, the rule application word determination unit 131 stores the word and application flag in the rule application word table 170. Is registered, the display mode determination unit 132 is supplied from the request processing unit 120 with reference to the stored contents of the rule application word table 170, the word importance storage unit 320, and the display rule storage unit 380 after the registration process. The display mode of each word in the constituent elements of the display screen is determined. Details of this processing will be described with reference to FIG. Here, a case will be described as an example where the rule application word determination unit 131 registers words and the like as shown in FIG. 33 in the rule application word table 170.

  First, the first word registered in the rule application word table 170 is acquired (step S801), and based on the application flag of the acquired word and the importance stored in the word importance storage unit 320 for the word. The display rule is extracted from the display rule storage unit 380 (see FIG. 27) (step S802). For example, since the application flag of the first word “3 cases” is “0”, the display rule corresponding to the application flag “0” is extracted. If the acquired word is “J League Kashima × Tokyo”, the application flag is “1” and the word importance is “50”, so the application flag is “1” and the importance range is “50 to less than 80”. Display rules to be extracted.

  When the display rule is extracted in this way, the display mode is corrected according to the display rule extracted for the word (step S803). When the target word is “3”, a correction is made such that the font is reduced by two, and when the target word is “J League Kashima × Tokyo”, the font is increased by two, In addition, the display mode is modified so that the display is bolded.

  When correction according to the display rule is performed as described above (there may be no correction), whether or not correction processing according to the display rule has been performed for all the words registered in the rule application word table 170. Is determined (step S804). If the processing has not been completed for all registered words, the next word is acquired from the rule application word table 170 (step S805), and the processing from step S802 is performed.

  On the other hand, when the process is completed for all registered words, the display mode determination process is terminated. By performing such processing, the contents stored in the recognition dictionary 160 among the contents to be displayed of the search result of the request processing unit 120, that is, among the words included in the constituent elements of the display screen, Accordingly, the display mode is corrected.

  As described above, among the words included in the constituent elements of the search result display screen by the display control unit 130, the speech-recognizable word is different from the other words, and the mode (font, The display content is controlled to be displayed in a font or the like. Then, a display screen of contents controlled by the display control unit 130 is displayed on the display unit 140. Here, an example of the display content of the display unit 140 is shown in FIG. As shown in the figure, among the words on the TV program search result display screen, words stored in the recognition dictionary 160, that is, words that can be recognized by voice, and a highly important sub-genre “baseball” or “Soccer” or the like (4 font sizes larger) is displayed the largest, and then the program name such as “J League Kashima × Tokyo” (font size 2 larger) is displayed larger.

  As described above, in the present embodiment, when the user makes a processing request, the request processing unit 120 processes the request, and the processing result (in the above example, the search result of the TV program) is displayed on the display unit 140. Is done. And the word which can be recognized with the said speech recognition apparatus 30 among the words contained in such a display screen can be displayed in the aspect different from another word (refer FIG. 34). As a result, the user who refers to the display can know the result of his request and can know a word that can be recognized by voice.

  Further, in the present embodiment, the importance level of the word is determined, and the aspect according to the importance level, that is, the emphasized form is displayed as the importance level increases, so that the user can recognize a recognizable word having a high importance level. Can be notified.

  Further, in the present embodiment, the importance is set to a high importance although it is highly likely that the importance will be input next in advance for each of various cases in consideration of the contents already input to the device. Therefore, it is possible to highlight and display the information that is likely to be input next when executing a certain process and notify the user. On the other hand, since words that are less likely to be input are not highlighted, words that are more likely to be input can be strongly impressed by the user. In the present embodiment, the search process is efficiently performed by determining the level of input possibility (importance level) in the search process according to the keyword attribute or the like input immediately before. The importance level can be set.

  Considering the program search process as an example as described above, the attributes already included in the search query after the previous utterance can be recognized, but they do not have an effect for the purpose of narrowing down the programs, so they need not be highlighted. Therefore, the importance is lowered, and as a result, the display mode is not emphasized. In addition, since programs searched for in a higher genre such as sports can be narrowed down effectively by sub-genre, by setting the importance of words belonging to the sub-genre of the higher-level genre high, it is efficient to achieve the processing purpose. A typical recognition target word is displayed with more emphasis. By making such a display, it is possible to quickly find and search the recognition target word, and efficiently achieve the requested processing purpose.

(Modification)
In addition, this invention is not limited to each embodiment mentioned above, The various deformation | transformation which is illustrated below is possible.

(Modification 1)
In the first embodiment described above, the display rule storage unit 180 has one type of display rule that is applied to words registered in the recognition dictionary 160. However, the rule that applies to registered words is stored in the display rule storage unit 180. A plurality of types may be prepared. For example, a rule may be provided to display the operation command word and the program attribute word in different forms, and the display mode of these words may be different.

(Modification 2)
Further, in each of the above-described embodiments, the font size and bolding are defined as the display rule. However, it is only necessary that the display can be displayed in different modes. For example, italic display, shading display, etc. You may make it use. Furthermore, any display mode may be used as long as it is distinguished from others, such as setting a color scheme or changing the size of the object (button or the like) on which the recognition word is displayed.

(Modification 3)
In the second embodiment described above, the display / reading determination unit 220 determines the reading most frequently used by the user in the past (reading with the highest frequency) when recognizing the recognition target word. Although the character string that changes the display mode (determines the display mode according to the display rule) is determined, the display mode may be changed based on other criteria.

  For example, the character string on the front side of the recognition target word may be determined as a reading method. Specifically, if the word to be recognized is “Monday Hysteria Theater“ The Condemnation Defense Attorney Series, Fumiaki Inokuma 5 ””, the three readings “Getsuyo Histeri Gekijo” and “ You may be made to select the reading “Getsuyo Histari Gejo” consisting only of the character string on the front side among “Z” and “Inokuma Fumiaki Faibu”.

  Further, only when the number of characters of the recognition target word is larger than a predetermined number (for example, 10 characters or more), a partial character string other than the entire word may be selected as a character string to be displayed differently. In this case, the rule application target character string may be determined on the basis of selecting the shortest character string among a plurality of readings of the recognition target word.

(Modification 4)
In addition, when there are a plurality of readings as in the second embodiment or the third modification, the reading mode is not limited to only one reading method and the display mode is changed only for the narrowed character string. In some cases, two or more readings may be displayed in such a manner that the user can be notified.

  For example, when the word “Monday hysterical theater“ accusal defense lawyer series, Inokuma Fumiaki 5 ”” that can be recognized by three readings as described above is displayed, the possible readings are recognized as shown in FIG. A range 401 may be designated in the upper and lower parts of the word character string. Further, it is possible to use a technique in which the display colors of these three readings are made different. For example, when a word not registered in the recognition dictionary 160 is displayed in black, “Monday Hystery Theater” is displayed in red, The “accusal defense attorney series” may be displayed in green, and the “Burmei Fumakuma 5” may be displayed in yellow.

  In the second embodiment, the character string that changes the display mode is determined based on the frequency of past reading by the user of the device. However, the frequency of past reading by the user of another device is determined. The character string that changes the display mode may be determined using the above. For example, among the three readings "Getsuyo Histeri Gekijo", "Kokuhatsubengo Ninshirizu", and "Inokuma Fumiaki Faibu", the reading most used by many users In the case of “Getsuyo Histari Gekijo”, this may be determined as a character string that changes the display mode. Information such as how a user of another device has read may be supplied to the device by using a network such as the Internet.

(Modification 5)
Further, in each of the above-described embodiments, the reading method for recognizing the recognition target word corresponds to all or a part of the word. However, by speaking the abbreviation of the recognition target word, the recognition target word is It may be recognized. In this case, in addition to reading the whole word, the reading of the abbreviation may be added to the reading column of the recognition dictionary 160 (260). For example, when the word to be recognized “Monday Hysteria Theater“ Prosecution of Defense Prosecutor Series / Fumiaki Inokuma 5 ”” is generally abbreviated as “Getsugeki” or the like, the recognition dictionary 160 (260) You should add “Getsugeki”.

  And in order to notify the user of such reading, it can be used for recognizing the word above or below the recognition target word “Monday Hysteria Theater“ Criminal Defense Attorney Series / Fumiaki Inokuma 5 ””. A display rule for displaying the abbreviation “Getsugeki” may be provided. In this way, according to the display rule, the display mode determination unit 132 corrects the display mode of the recognition target word “Monday hysterical theater“ accusal defense lawyer series: Fumaaki Inokuma 5 ”” and displays the display as shown in FIG. Is made.

(Modification 6)
In each of the above-described embodiments, the request processing unit 120 performs a search process based on a search keyword or the like input by the user. However, the request processing unit 120 performs other processes in response to a user request. The present invention can be applied even to an apparatus that performs the same. For example, the present invention can be applied to a voice recognition device mounted on an air conditioner device.

  More specifically, words such as “operation start”, “operation stop”, “large air flow”, “setting temperature up”, “setting temperature down”, etc. are registered in the recognition dictionary 160, and the air conditioner of the user is registered. In response to a request to the device (operation, etc.), the processing result for the request is displayed on the display panel etc. (for example, when a request to start operation is displayed, a message such as starting operation) is displayed. Of the words included in the screen, recognizable words may be displayed in a manner different from other words in accordance with display rules.

  In addition, when the present invention is applied to the above-described air conditioner device or the like, when the function of determining the importance of a word is provided as in the third embodiment, the point in time at the time of determining the importance of the word The operating condition of the air conditioner device in the above may be used as one element for determining the importance. For example, while driving, the word "start driving" is unlikely to be input again, so the word is set to be less important, while words related to air volume adjustment, temperature adjustment, etc. during driving The operation status may be reflected in the determination of the importance level such that the importance level is set high.

(Modification 7)
In addition, the voice recognition device according to each of the above embodiments may be mounted on a recorder device that records a moving image such as a television program. Here, FIG. 37 shows a schematic configuration example of a recorder apparatus 400 incorporating the speech recognition apparatus 10 according to the first embodiment. As shown in the figure, the recorder device 400 includes the voice recognition device 10 having the above-described configuration, a recorder unit 410, and a tuner unit 420.

  The tuner unit 420 selects a predetermined television channel according to a user instruction or the like, and receives a moving image of the selected channel. When viewing a television program to be broadcast, the moving image received by the tuner unit 420 is supplied to the display unit 140, and the television program or the like is displayed.

  The recorder unit 410 records the moving image (such as a television program) received by the tuner unit 420 as described above on a recording medium. For example, the received moving image is not recorded as it is, but is compressed by a compression method such as MPEG (Moving Pictures Experts Group) -2, and the compressed moving image data is recorded. The recording medium may be an HD (Hard Disk) or the like, or a general-purpose recording medium (DVD-RAM (Digital Versatile Disc-RAM) or the like).

  In the recorder apparatus 400 having such a configuration, the user can search for a program or the like to be recorded. When performing such a search, the utterance content is recognized by the speech recognition apparatus 10 by uttering a word about the attributes (genre, name, broadcast date and time, performer) of the program to be recorded, and is described in the above embodiment. As described above, the request processing unit 120 performs a search using the utterance content as a keyword.

  Then, the search result is displayed on the display unit 140. At this time, the display control unit 130 displays words that can be recognized by voice in a different mode (highlighted mode) from other characters. It is possible to recognize which words can be input when trying to input the words for the voice. Then, the search contents are uttered to search for a TV program, and when a desired program is searched, an instruction is issued to record it. Thereby, the request processing unit 120 sends a recording instruction including information for identifying the searched program to the recorder unit 410, and the recorder unit 410 performs recording according to the instruction of the request processing unit 120.

(Modification 8)
In each of the above-described embodiments, the case where the present invention is applied to the speech recognition apparatus has been described. However, as the display control apparatus including the display control unit 130, the rule application word table 170, and the display rule storage unit 180, the speech You may make it distribute | circulate by manufacturing, selling, etc. in the aspect isolate | separated from the speech recognition apparatus provided with the input part 100, the speech recognition part 110, a dictionary.

(Modification 9)
The display mode control process performed in each of the above-described embodiments may be performed by a dedicated hardware circuit, or may be configured so that the process is performed when the CPU operates according to a program. May be. Further, a program for causing the computer to execute such processing may be provided to the user via a communication line such as the Internet, or the program may be provided as a CD-ROM (Compact Disc-Read Only Memory). It may be recorded on a computer-readable recording medium and provided to the user.

  As described above, the speech recognition device, display control device, recorder device, display method, and program according to the present invention are particularly suitable for a device having a function of performing a search process on a keyword input by speech recognition.

It is a block diagram which shows the structure of the speech recognition apparatus concerning the 1st Embodiment of this invention. It is a figure which shows an example of the registration content of the recognition dictionary which is a component of the said speech recognition apparatus. It is a figure which shows an example of the program database memorize | stored in the information processing part for request processes which is a component of the said speech recognition apparatus. It is a figure which shows an example of the registration content of the rule application word table which is a component of the said speech recognition apparatus. It is a figure which shows an example of the display rule memorize | stored in the display rule memory | storage part which is a component of the said speech recognition apparatus. It is a flowchart which shows an example of the process sequence of the request | requirement process part which is a component of the said speech recognition apparatus. It is a figure which shows an example of the template utilized when the said request | requirement process part produces the screen for displaying a process result. It is a figure which shows an example of the display screen which shows the process result of the said request process. It is a flowchart which shows the procedure of the process performed by the rule application word determination part which is a component of the said speech recognition apparatus. It is a flowchart which shows the procedure of the process performed by the display mode determination part which is a component of the said speech recognition apparatus. It is a figure which shows an example of the display screen corrected by the said display mode determination part. It is a block diagram which shows the structure of the speech recognition apparatus concerning the 2nd Embodiment of this invention. It is a figure which shows an example of the registration content of the recognition dictionary of the speech recognition apparatus in 2nd Embodiment. It is a figure which shows an example of the memory content of the reading history memory | storage part of the speech recognition apparatus in 2nd Embodiment. It is a figure which shows an example of the registration content of the rule application word table in 2nd Embodiment. It is a figure which shows an example of the program database memorize | stored in the information processing part for request | requirement processing in 2nd Embodiment. It is a flowchart which shows the procedure of the process performed by the reading frequency management part which is a component of the speech recognition apparatus in 2nd Embodiment. It is a flowchart which shows the procedure of the process performed by the display reading determination part which is a component of the speech recognition apparatus in 2nd Embodiment. It is a figure which shows an example of the display screen produced by the request | requirement process part of the speech recognition apparatus in 2nd Embodiment. It is a figure which shows an example of the registration content of the said rule application word table in which the rule application object character string was written by the said display stagnation way determination part. It is a flowchart which shows the procedure of the process performed by the display mode determination part in 2nd Embodiment. It is a figure which shows an example of the display rule memorize | stored in the display rule memory | storage part in 2nd Embodiment. It is a figure which shows an example of the display screen corrected by the display mode determination part in 2nd Embodiment. It is a block diagram which shows the structure of the speech recognition apparatus concerning the 3rd Embodiment of this invention. It is a figure which shows an example of the importance determination rule memorize | stored in the importance determination rule memory | storage part which is a component of the speech recognition apparatus in 3rd Embodiment. It is a figure which shows an example of the memory content of the word importance memory | storage part which is a component of the speech recognition apparatus in 3rd Embodiment. It is a figure which shows an example of the display rule memorize | stored in the display rule memory | storage part in 3rd Embodiment. It is a figure which shows an example of the screen for displaying the processing result produced by the request | requirement process part in 3rd Embodiment. It is a figure which shows an example of the template utilized when the said request process part produces a display screen. It is a flowchart which shows the procedure of the process performed by the word importance determination part which is a component of the speech recognition apparatus in 3rd Embodiment. It is a figure which shows an example of the registration content of the recognition dictionary in 3rd Embodiment. It is a flowchart which shows the procedure of the process performed by the display mode determination part in 3rd Embodiment. It is a figure which shows an example of the registration content of the rule application word table in 3rd Embodiment. It is a figure which shows an example of the display screen corrected by the display mode determination part in 3rd Embodiment. It is a figure which shows an example of the display mode of the word registered into the recognition dictionary displayed on the display part of the modification of 2nd Embodiment. It is a figure which shows an example of the display mode of the word registered into the recognition dictionary displayed on the display part of the modification of each embodiment. It is a block diagram which shows the structure of the recorder apparatus provided with the speech recognition apparatus concerning 1st Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Speech recognition apparatus 20 Speech recognition apparatus 30 Speech recognition apparatus 100 Speech input part 110 Speech recognition part 120 Request processing part 130 Display control part 131 Rule application word determination part 132 Display mode determination part 140 Display part 150 Acoustic dictionary 160 Recognition dictionary 170 Rule Application word table 180 Display rule storage unit 190 Request processing information storage unit 210 Reading frequency management unit 220 Display reading determination unit 230 Reading history storage unit 260 Recognition dictionary 270 Rule application word table 310 Word importance determination unit 320 Word importance storage unit 330 Importance Determination Rule Storage Unit 380 Display Rule Storage Unit 400 Recorder Device 401 Frame 410 Recorder Unit 420 Tuner Unit

Claims (12)

A recognition dictionary that stores multiple words that are subject to speech recognition and how to read these words;
Speech recognition means for performing speech recognition processing with reference to the recognition dictionary for input speech;
Display rule storage means for storing a display rule defining a display mode of words stored in the recognition dictionary;
When displaying the request processing result for the input request on the display screen, if a word stored in the recognition dictionary is included in the display screen, the displayed rule is stored in the display rule storage unit. Display control means for determining a display mode according to stored display rules;
A speech recognition apparatus comprising: A recognition dictionary that stores multiple words that are subject to speech recognition and how to read these words;
Speech recognition means for performing speech recognition processing with reference to the recognition dictionary for input speech;
Display rule storage means for storing a display rule defining a display mode of words stored in the recognition dictionary;
Request processing means for processing a request based on the voice recognized by the voice recognition means;
When displaying the request processing result by the request processing means for the input request on the display screen, if the display screen includes a word stored in the recognition dictionary, the word included Display control means for determining a display mode according to the display rules stored in the display rule storage means;
A speech recognition apparatus comprising: For words in which a plurality of types of readings are stored in the recognition dictionary, further comprising a display reading-compatible character string determination means for determining a character string based on a predetermined criterion,
The voice recognition apparatus according to claim 1, wherein the display control unit determines a display mode of the character string determined by the display reading correspondence character string determination unit according to the display rule. For words that can be read in a plurality of types in the recognition dictionary, it further comprises reading frequency management means for recording the frequency of reading input when recognized by the voice recognition means,
The display reading determination means displays a character string corresponding to the reading with the highest reading frequency recorded by the reading frequency management means for a word in which a plurality of types of readings are stored in the recognition dictionary according to the display rule. The speech recognition apparatus according to claim 3, wherein the mode is determined as a character string to be determined. The display / reading determination means, for words whose number of characters constituting the word stored in the recognition dictionary is smaller than a predetermined value, how to read the whole word as a character string whose display mode should be determined according to the display rule The speech recognition apparatus according to claim 3 or 4, wherein the voice recognition apparatus is determined. Further comprising importance determining means for determining the importance of the words stored in the recognition dictionary;
The voice recognition apparatus according to claim 1, wherein the display control unit determines a display mode based on the display rule and the importance determined by the importance determination unit. . The importance level determination means determines the importance level for words included in the result displayed on the display screen as a result of the request processing means for the input request based on the content of the request already input. The speech recognition apparatus according to claim 6. The speech recognition apparatus according to claim 6, wherein the importance level determination unit determines the importance level based on an operation state of the request processing unit. Speech recognition comprising a recognition dictionary for storing a plurality of words to be speech-recognized and how to read these words, and speech recognition means for performing speech recognition processing on the input speech by referring to the recognition dictionary A display control device for displaying words that can be recognized by the device,
Display control means for controlling the display contents when displaying the request processing result for the input request on the display screen;
A display rule storage means for storing a display rule for defining a display mode of words stored in the recognition dictionary;
The display control means, when the word to be stored in the recognition dictionary is included in the result to be displayed on the display screen, the display stored in the display rule storage means for the included word A display control device that determines a display mode according to a rule. A recorder device that performs processing for storing an image,
A recognition dictionary that stores multiple words that are subject to speech recognition and how to read these words;
Speech recognition means for performing speech recognition processing with reference to the recognition dictionary for input speech;
Request processing means for processing a request based on the voice recognized by the voice recognition means;
Display control means for controlling the display contents when the result of the request processing means for the request is displayed on a display screen;
A display rule storage means for storing a display rule for defining a display mode of words stored in the recognition dictionary;
The display control means, when the word to be stored in the recognition dictionary is included in the result to be displayed on the display screen, the display stored in the display rule storage means for the included word A display device that determines a display mode according to a rule. Speech recognition comprising a recognition dictionary for storing a plurality of words to be speech-recognized and how to read these words, and speech recognition means for performing speech recognition processing on the input speech by referring to the recognition dictionary A method for displaying words that the device can recognize,
When the request processing result for the input request is displayed on the display screen, if the word to be stored in the recognition dictionary is included in the result to be displayed on the display screen, the included word A display method characterized by determining a display mode according to a predetermined display rule. Computer
Included when the request processing result for the input request is displayed on the display screen, the result to be displayed on the display screen includes a word stored in the recognition dictionary used for the speech recognition processing A program that functions as a display control means for determining a display mode according to a display rule determined in advance for a word that is stored.

Images