JP2011170103A - Advertisement display system, advertisement display method, and advertisement display program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve an advertisement effect by naturally presenting an advertisement word to a user who utilizes voice recognition processing. <P>SOLUTION: An advertisement display system 100 includes: a voice recognition unit 102 which calculates the sound score, the language score and the total score, of a word to be the candidate of a voice recognition result of input voice data, to select a word having high total score, as a voice recognition result; an advertisement word storage unit 120 for storing the advertisement word in which an advertisement rate is paid as a counter value for the word to be easily selected as the voice recognition result; and an output adjustment unit 104 which performs adjustment so that the advertisement word is selected as the voice recognition result of the voice data when the similarity degree of the advertisement word with the voice data is within a predetermined similarity allowable range, and which displays and outputs the adjusted result as the voice recognition result of the voice data. Moreover, the similarity allowable range is set wider, as the rank of the advertisement rate of the advertisement word is higher. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an advertisement display system, an advertisement display method, and an advertisement display program.

  Patent Document 1 (Japanese Patent Application Laid-Open No. 11-133994) discloses a voice input device that selects an optimum one from a plurality of recognition result candidates obtained for an input voice and obtains a recognition result for the input voice. An adaptive score storage unit that stores adaptive scores of a plurality of words that can be recognition result candidates, and a selection result that is stored in the adaptive score storage unit when a recognition result is selected from a plurality of recognition result candidates obtained for input speech. Means for selecting the recognition result in consideration of the adaptive score of each recognition result candidate, and correcting the recognition result selected by the means according to the user's instruction, and the corrected recognition result is selected by the means A voice input device is described that includes a user correction unit that updates the contents of the adaptive score storage unit so as to facilitate.

  Patent Document 2 (Japanese Patent Application Laid-Open No. 2008-170820) discloses a means for storing a word and content in association with each other, a word extracting means for extracting a word from conversation speech, and a word extracted by the word extracting means. A content providing system including a content reading unit that reads the content stored in association with the content transmission unit that transmits the read content to a content reproduction unit is described. Thereby, it is supposed that the advertisement etc. which should be displayed can be selected based on a speaker's emotion.

JP-A-11-133994 JP 2008-170820 A

  However, in the conventional advertisement display method, the content associated with the word is presented to the user separately from the speech recognition result, as in the technique described in Patent Document 2, for example. Therefore, for the user, the content is displayed even though the user did not try to display the content. For example, when the content is an advertisement, the user clearly understands that the content is an advertisement. It has become unnatural.

  An object of the present invention is to provide an advertisement display system, an advertisement display method, and an advertisement display program that solve the above-described problem of unnaturalness when an advertisement is presented to a user.

According to the present invention,
Based on the acoustic model and the language model, the acoustic score of the words that are candidates for the speech recognition result for the input speech data, the language score, and the total score based on the acoustic score and the language score are calculated, and the total score is high. Speech recognition means for selecting words as speech recognition results;
Including an advertising word storage means for storing an advertising word for which an advertising fee is paid as a consideration for facilitating selection as a speech recognition result, and the similarity between the advertising word and the voice data is within a predetermined similarity allowable range Adjusting so that the advertisement word is selected as the voice recognition result of the voice data, and output adjusting means for displaying and outputting the adjusted result as the voice recognition result of the voice data;
Including
There is provided an advertisement display system that is set such that the similarity allowable range becomes wider as the rank of the advertisement fee of the advertisement word is higher.

According to the present invention,
An advertisement display method using a computer system including an advertisement word storage means for storing an advertisement word for which an advertisement fee is paid as a price for facilitating selection as a speech recognition result,
Based on the acoustic model and the language model, an acoustic score, a language score, and a total score based on the acoustic score and the language score are calculated for a word that is a speech recognition result candidate for the input speech data. A speech recognition step for selecting high words as speech recognition results;
When the similarity between the advertisement word and the voice data is within a predetermined similarity tolerance range, the advertisement word is adjusted to be selected as a voice recognition result of the voice data, and the adjusted result is the voice data. Output adjustment step to display and output as a voice recognition result of
Including
There is provided an advertisement display method that is set such that the similarity allowable range becomes wider as the advertisement charge rank of the advertisement word is higher.

According to the present invention,
Computer
Based on the acoustic model and the language model, an acoustic score, a language score, and a total score based on the acoustic score and the language score are calculated for a word that is a speech recognition result candidate for the input speech data. Speech recognition means for selecting high words as speech recognition results,
Including an advertising word storage means for storing an advertising word for which an advertising fee is paid as a consideration for facilitating selection as a speech recognition result, and the similarity between the advertising word and the voice data is within a predetermined similarity allowable range Adjusting the output so that the advertisement word is selected as the voice recognition result of the voice data, and displaying and outputting the adjusted result as the voice recognition result of the voice data;
A program that functions as
There is provided an advertisement display program which is set such that the similarity allowable range becomes wider as the rank of the advertisement fee of the advertisement word is higher.

  It should be noted that any combination of the above-described constituent elements and a conversion of the expression of the present invention between a method, an apparatus, a system, a recording medium, a computer program, etc. are also effective as an aspect of the present invention.

  ADVANTAGE OF THE INVENTION According to this invention, an advertising word can be shown to a user using a speech recognition process naturally, and an advertising effect can be heightened.

It is a block diagram which shows an example of a structure of the advertisement display system in embodiment of this invention. It is a figure which shows an example of a structure of the advertisement word memory | storage part in embodiment of this invention. It is a flowchart which shows the process sequence of the advertisement display system in embodiment of this invention. It is a schematic diagram which shows the process sequence of the score adjustment part in embodiment of this invention. It is a schematic diagram which shows the process sequence of the score adjustment part in embodiment of this invention. It is a figure which shows the other example of a structure of the advertisement word memory | storage part in embodiment of this invention. It is a block diagram which shows an example of a structure of the advertisement display system in embodiment of this invention. It is a figure which shows an example of a structure of the advertisement word memory | storage part in embodiment of this invention. It is a figure which shows the other example of a structure of the advertisement word memory | storage part in embodiment of this invention. It is a block diagram which shows an example of a structure of the advertisement display system in embodiment of this invention. It is a figure which shows an example of a structure of the advertisement word memory | storage part in embodiment of this invention. It is a block diagram which shows the other example of a structure of the advertisement display system in embodiment of this invention. It is a flowchart which shows the process sequence of the advertisement display system in embodiment of this invention. It is a block diagram which shows the network structure containing the advertisement display system in embodiment of this invention. It is a block diagram which shows the other example of a structure of the advertisement display system in embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In all the drawings, the same constituent elements are denoted by the same reference numerals, and the description thereof is omitted as appropriate.

In the following embodiment, the advertisement display system includes:
Based on the acoustic model and the language model, the acoustic score of the words that are candidates for the speech recognition result for the input speech data, the language score, and the total score based on the acoustic score and the language score are calculated, and the total score is high. Speech recognition means for selecting words as speech recognition results;
Including an advertising word storage means for storing an advertising word for which an advertising fee is paid as a consideration for facilitating selection as a speech recognition result, and the similarity between the advertising word and voice data is within a predetermined similarity allowable range, Adjusting so that the advertisement word is selected as the voice recognition result of the voice data, and output adjusting means for displaying and outputting the adjusted result as the voice recognition result of the voice data;
Including
The similarity allowable range is set to be wider as the rank of the advertisement fee of the advertisement word is higher.

  In the following embodiment, the advertisement display system is configured such that when the sound of the advertisement word and the input voice data is somewhat similar, the total score of the original speech recognition processing of the advertisement word is higher than the total score of other words. Even if it is low, the advertisement word is displayed and output as a voice recognition result. Here, to what extent the sound of the advertising word and the input voice data is similar can be defined by setting a similar allowable range. The degree of sound similarity between the advertisement word and the input voice data can be determined based on the total score or acoustic score of the result of the voice recognition process.

(First embodiment)
FIG. 1 is a block diagram illustrating an example of a configuration of an advertisement display system in the present embodiment.
The advertisement display system 100 includes a speech recognition unit 102 (speech recognition unit), an output adjustment unit 104 (output adjustment unit), an acoustic model storage unit 110, and a language model storage unit 112.

  The acoustic model storage unit 110 stores an acoustic model. The language model storage unit 112 stores a language model. These acoustic models and language models can be generally used.

  The speech recognition unit 102 is based on the acoustic model stored in the acoustic model storage unit 110 and the language model stored in the language model storage unit 112, and the acoustic score of words that are candidates for speech recognition results for the input speech data, A language score and a total score based on the acoustic score and the language score are calculated. The voice recognition unit 102 selects a word having a high total score as a voice recognition result of voice data that has been input. In the present embodiment, the speech recognition unit 102 can select a plurality of words having a high total score in association with the total score in descending order of the total score.

  The output adjustment unit 104 performs adjustment so that the advertisement word is selected as the voice recognition result of the voice data when the similarity between the advertisement word and the input voice data is within a predetermined similarity tolerance range. The output adjustment unit 104 displays and outputs the adjusted result as a voice recognition result of the voice data. Here, the output display means output in a form displayed on a display or the like of another terminal connected via a network, as will be described later.

  Specific examples are shown below. In the present embodiment, the output adjustment unit 104 includes a score adjustment unit 106 (score adjustment unit), a display output unit 108 (display output unit), and an advertisement word storage unit 120 (advertisement word storage unit).

  The advertising word storage unit 120 stores an advertising word for which an advertising fee is paid as a consideration for facilitating selection as a voice recognition result. In the present embodiment, the advertising word storage unit 120 stores a correction value of a score that becomes a higher value as the advertising fee rank of the advertising word is higher in association with the advertising word as a parameter for defining the similarity allowable range. . As a result, the similar allowable range is set to be wider as the advertisement charge rank of the advertisement word is higher.

  For the advertising word, the score adjustment unit 106 performs adjustment to add the score correction value stored in the advertising word storage unit 120 to the total score calculated by the voice recognition unit 102.

FIG. 2 is a diagram illustrating an example of the configuration of the advertising word storage unit 120.
The advertising word storage unit 120 includes an advertising word column, an advertising fee column, and a score correction value column. In the advertisement word column, an advertisement word that the advertiser wants to advertise, such as “NEC” or “Papello” is stored.

  In the advertisement fee column, information indicating the advertisement fee of each advertisement word such as “20 yen / output”, “10 yen / output”, etc. is stored. “20 yen / output” means that the advertisement word is not originally displayed and output as a result of the normal speech recognition process, but the advertisement word is displayed and output by the adjustment process of the advertisement display system 100 in the present embodiment. This means that an advertising fee of 20 yen per output is imposed.

  In the score correction value column, for example, score correction values such as “20” and “10” are stored. In the present embodiment, the score correction value is set as a parameter that defines the similar allowable range. Here, the correction value of the score can be a correction value for the total score.

  Hereinafter, a method for setting a score correction value will be described. The correction value of the score is, for example, human if the difference between the total score of the highest word having the highest total score for the input voice data and the total score of the advertising word for the voice data is within the range. The upper limit value can be determined in consideration of whether they feel similar to each other. If the correction value of the score is too large, the advertisement word is selected as the voice recognition result of the voice data even if the voice data input to the voice recognition unit 102 and the advertisement word are completely different as sounds. turn into. However, if an advertisement word whose voice is completely different from the input voice data is displayed and output as a voice recognition result, it becomes unnatural for the user. Therefore, in the present embodiment, the upper limit value of the score correction value is determined as described above, and the score correction value is controlled to be equal to or lower than the upper limit value.

  Further, in the present embodiment, the score correction value can be set to be higher as the rank of the advertisement fee of the advertisement word is higher. For example, the advertising word “NEC” whose unit price of the advertising fee is “20 yen / output” has a higher advertising fee rank than the advertising word “Papello” whose unit price of the advertising fee is “10 yen / output”. Therefore, in the advertising word storage unit 120, the score correction value of the advertising word “NEC” is set to be “20”, which is higher than the correction value “10” of the advertising word “Papello”. Here, the score correction value is set according to each unit price of the advertising fee. For example, the score correction value is classified according to the unit price of the advertising fee, and the score correction value is set according to each rank. You can also. For example, if the unit price of the advertising fee is “6 yen / output” or more and “10 yen / output” or less, “C (low rank)”, and the unit price of the advertising fee is “11 yen / output” or more and “15 yen / output” or less. “B (medium rank)”, and “A (high rank)” when the unit price of the advertising fee is “16 yen / output” or more and “20 yen / output” or less, a score correction value is set for each rank. You can also

  In the present embodiment, the correction value of the score of each rank can be set in accordance with the rank of the advertisement fee of the advertisement word within the range not exceeding the upper limit of the correction value of the score. For example, if the difference between the total score of the highest word having the highest total score for the input voice data and the total score of the advertising word for the voice data is within 30, it is determined that these words are similar. In such a case, the upper limit of the score correction value can be set to 30, and the score correction value can be set to 30, 25, 20,.

  In the present embodiment, an example is shown in which an advertising fee column is provided in the advertising word storage unit 120, but the advertising word storage unit 120 may be configured not to include an advertising fee column.

  Returning to FIG. 1, the display output unit 108 displays and outputs a word having a high total score as a voice recognition result of the input voice data based on the score adjusted by the score adjustment unit 106. In the present embodiment, the display output unit 108 can display and output only one word having the highest total score.

FIG. 14 is a block diagram showing a network structure including the advertisement display system 100 in the present embodiment.
This network structure includes an advertisement display system 100 and a user terminal device 200 connected to the advertisement display system 100 via a network 150 such as the Internet. The user terminal device 200 can be a user's PC or the like. The user terminal device 200 can be configured by a personal computer, for example. The user terminal device 200 can be configured to include voice input means such as a microphone and display means such as a display. When voice data is input via the voice input unit of the user terminal 200, the voice data is input to the voice recognition unit 102 (see FIG. 1) of the advertisement display system 100 via the network 150. When the display output unit 108 of the advertisement display system 100 outputs a speech recognition result, the recognition result is input to the user terminal device 200 via the network 150 and is displayed and output on the display unit of the user terminal device 200. Note that the voice data of the user of the user terminal 200 is input to the voice recognition unit 102 of the advertisement display system 100 via a network other than the network 150 such as a telephone line if the correspondence with the user terminal 200 is achieved. It can also be configured.

  Next, in this embodiment, a procedure from when voice data is input to the advertisement display system 100 to when a voice recognition result is displayed and output will be described. FIG. 3 is a flowchart showing a processing procedure of the advertisement display system 100 in the present embodiment.

  When voice data is input (step S100), the voice recognition unit 102 performs voice recognition processing (step S102). This process can be a normal voice recognition process. Specifically, the speech recognition unit 102 calculates each score of words that are candidates for speech recognition results.

  Subsequently, the score adjustment unit 106 of the output adjustment unit 104 adjusts the score using the score correction value stored in the advertisement word storage unit 120 for the advertisement word with respect to the speech recognition result by the speech recognition unit 102. Is performed (step S104). The display output unit 108 outputs a word having a high score as a speech recognition result based on the score after the score adjustment by the score adjustment unit 106 is performed (step S106).

  The processing procedure of the output adjustment unit 104 in the present embodiment will be described with reference to FIGS. Here, the display output unit 108 displays and outputs only one word having the highest total score as a speech recognition result.

  As shown in FIG. 4A, for example, it is assumed that the input voice is “I went to life”. Among these, as a speech recognition result of a word corresponding to “enichi”, “enjoyment day”,..., “NEC”... Are assumed as candidate words (FIG. 4B). Further, it is assumed that the total score of the result of the normal speech recognition processing by the speech recognition unit 102 is “Family Day” = “30”, “NEC” = “16”. In FIG. 4B, candidates are displayed in descending order of the total score. In this case, in the normal speech recognition process, the “day of the festival” having the highest total score is displayed and output as the speech recognition result as a result of the speech recognition process.

  In the present embodiment, the result of the speech recognition process of the speech recognition unit 102 is not output as it is, but the result is output after the score adjustment unit 106 adjusts the score for the advertising word. The score adjustment unit 106 refers to the advertisement word storage unit 120 and adds a correction value of the score to the total score as a result of the speech recognition processing by the speech recognition unit 102 for the advertisement word. As shown in FIG. 2, “NEC” is stored as an advertising word in the advertising word storage unit 120, and the score correction value is “20”. Therefore, when the correction value “20” of the score is added to the total score “16” of “NEC” shown in FIG. 4B, the total score of “NEC” after adjustment becomes “36”. As a result, the total score of “NEC” becomes larger than the total score “30” of “Family Day”, and becomes the largest (FIG. 4C).

  Therefore, the display output unit 108 selects “NEC” having the highest total score based on the total score after the adjustment process as the voice recognition result of the input voice data “Enichi”, and displays “NEC” as the voice recognition result. “I went” is displayed and output (FIG. 4D).

  On the other hand, as shown in FIG. 5 (a), for example, it is assumed that the input voice is “I went to Eigai”. Among these, as a speech recognition result of a word corresponding to “Eiga”, “movie”,..., “NEC”..., Etc. are cited as candidate words (FIG. 5B). Further, it is assumed that the total score of the result of the normal voice recognition processing by the voice recognition unit 102 is “movie” = “30”, “NEC” = “8”. Also in FIG. 5B, candidates are displayed in descending order of the total score.

  As described with reference to FIG. 4, the score adjustment unit 106 refers to the advertisement word storage unit 120 and corrects the score to the total score as a result of the speech recognition processing by the speech recognition unit 102 for the advertisement word. Add a value. In this case, when the correction value “20” of the score is added to the total score “8” of “NEC” shown in FIG. 5B, the adjusted score of “NEC” becomes “28”. In this example, after adjusting the score of “NEC”, the score “30” of “movie” is larger (FIG. 5C). Therefore, in this example, the score adjustment unit 106 selects “movie” having the highest score based on the score after the adjustment processing as the voice recognition result of the input voice data “Eiga”, and the voice recognition result “I went to the movie” or the like is output (FIG. 5D).

  In the present embodiment, the advertisement display system 100 adjusts so that the advertisement word is selected as the voice recognition result of the voice data when the sound of the advertisement word and the input voice data is similar. Therefore, by adjusting the language score rather than adjusting the acoustic score indicating the degree of similarity of the sound itself, when the sound is similar to the input voice data, the advertising word is easily displayed and output. Can do.

  Hereinafter, as another example, a case where the score correction value is for the language score will be described. In the score correction value column of the acoustic model storage unit 120, a correction value for the language score is stored. In this case, for the advertising word, the score adjustment unit 106 performs adjustment to add the score correction value stored in the advertising word storage unit 120 to the language score calculated by the speech recognition unit 102. When the score adjustment unit 106 adds a correction value to the language score of the advertising word, the voice recognition unit 102 can calculate the total score based on the language score and the acoustic score to which the correction value is added.

  In this case, the correction value of the language score is, for example, within the range of the difference between the total score of the highest word having the highest total score with respect to the input voice data and the total score of the advertising word with respect to the voice data Then, the upper limit value of the correction value of the total score is determined in consideration of whether the person feels similar, and the upper limit value of the correction value of the language score corresponding to the upper limit value can be determined.

  As described above, according to the advertisement display system 100 in the present embodiment, even when the original total score of the advertisement word is lower than the total score of other words, the similarity between the advertisement word and the voice data is predetermined. If it is within the similar permissible range, the advertisement word is displayed and output as the voice recognition result of the voice data. In the present embodiment, a score correction value is set as a parameter that defines the similar allowable range. The score correction value is similar if the difference between the total score of the highest word with the highest total score for the input voice data and the total score of the advertising words for the voice data is within the range. The upper limit is determined in consideration of whether it can be determined. Therefore, the advertising word is displayed and output only when the advertising word is similar to the input voice data, so that the advertising word can be presented to the user in a natural manner to enhance the advertising effect. Can do.

  For example, in the example described with reference to FIG. 4 and FIG. 5, if the input voice data is “Eniichi”, the advertisement word “NEC” is displayed as a voice recognition result, assuming that it is similar to the voice data. However, when the input voice data is “Eiga”, the advertising word “NEC” is not displayed and output as a voice recognition result. Since “enichi” and “NEC” have similar sounds, the user may feel that “NEC” is displayed as a misrecognition result and may not notice that “NEC” is an advertising word. As a result, the word “NEC” can be caught naturally by the user, and the word can be impressed by the user.

  FIG. 2 shows an example in which the correction value of the score is stored in the advertising word storage unit 120. However, in the initial setting, the advertising word storage unit 120 associates the advertising word with the advertising word. Only the information indicating the fee may be stored.

  This configuration is shown in FIG. In this case, the output adjustment unit 104 sets the score correction value based on the information indicating the advertisement fee of the advertisement word stored in the advertisement word storage unit 120 so that the higher the rank of the advertisement fee of the advertisement word, the higher the value. It can be configured to have a correction value setting means for setting. In the present embodiment, the score adjustment unit 106 can also have a function of a correction value setting unit. The score adjustment unit 106 can accept an input of an upper limit value of the score correction value, and can set a score correction value in accordance with the rank of the advertisement fee of the advertisement word within a range equal to or lower than the upper limit value.

  As shown in FIG. 6B, the score adjustment unit 106 can store the set score correction value in association with each advertisement word in the advertisement word storage unit 120. In another example, the score adjustment unit 106 may store a score correction value in association with each advertisement word in a storage unit different from the advertisement word storage unit 120.

  The subsequent processing can be the same as that of the advertisement display system 100 described with reference to FIGS. Even with such a configuration, the same effect as the advertisement display system 100 described with reference to FIGS. 1 to 5 can be obtained.

(Second Embodiment)
FIG. 7 is a block diagram showing an example of the configuration of the advertisement display system in the present embodiment.
This embodiment is different from the first embodiment in that the output adjustment unit 104 includes a similarity determination unit 130 instead of the score adjustment unit 106 shown in FIG. Further, in the present embodiment, the content that the advertising word storage unit 120 stores as a parameter that defines the similar allowable range is also different from that of the first embodiment.

  In the present embodiment, the advertisement display system 100 displays and outputs the advertisement word as a result instead of the voice recognition processing result of the actual voice data when the sound of the advertisement word and the input voice data is similar. Like that. Therefore, whether or not the advertising word is similar to the input voice data can be determined based on the acoustic score or the total score, not the language score.

  As an example, the advertising word storage unit 120 stores, as a parameter for defining the similarity allowable range, a score allowable range that becomes higher as the advertising charge rank of the advertising word is higher in association with the advertising word. The similarity determination unit 130 calculates a score between the acoustic score or the total score of the highest word having the highest acoustic score or the total score calculated by the voice recognition unit 102 and the acoustic score or the total score of the advertisement word for the audio data. If the difference is within the allowable range of the score, and if it is within the allowable range, it is determined that the similarity between the advertisement word and the audio data is within the allowable range of similarity. When the similarity determination unit 130 determines that the similarity between the advertisement word and the voice data is within the similarity allowable range, the display output unit 108 displays and outputs the advertisement word as a voice recognition result of the voice data. In other words, the wider the score tolerance range, the higher the possibility that it is determined that the advertising word is similar to the input voice data, and the advertising word is easily displayed and output.

FIG. 8 is a diagram showing an example of the configuration of the advertising word storage unit 120 in the present embodiment.
The advertising word storage unit 120 includes an advertising word column, an advertising fee column, and a score tolerance range column. The advertising word column and the advertising fee column can be the same as the configuration shown in FIG. In the score tolerance range column, for example, a score tolerance range such as “20” or “10” is stored. The allowable range of the score can be either an allowable range for the total score or an allowable range for the acoustic score depending on the processing of the similarity determining unit 130.

  For example, as described with reference to FIG. 4 in the first embodiment, it is assumed that the input voice is “I went to life”. In addition, as a speech recognition result of a word corresponding to “enichi”, it is assumed that “female day”,..., “NEC”... Are candidate words (FIG. 4B). Further, it is assumed that the total score of the result of the normal speech recognition processing by the speech recognition unit 102 is “Family Day” = “30”, “NEC” = “16”. Here, the difference between the total score “30” of “Fair Day” with the highest total score and the total score “16” of the advertising word “NEC” is “14”, and the advertising word “NEC” shown in FIG. Is within the allowable range of “20”. For this reason, the similarity determination unit 130 determines that the similarity between the advertising word “NEC” and the voice data is within the allowable range of similarity, and the display output unit 108 indicates “I went to NEC” or the like as the voice recognition result. Output.

  As another example, the advertising word storage unit 120 stores, as a parameter that defines the similarity allowable range, an allowable range having a higher value as the advertising fee rank of the advertising word is higher in association with the advertising word. The similarity determination unit 130 determines whether or not the rank of the advertising word is within an allowable range when the speech data is arranged as a speech recognition result candidate in descending order of the acoustic score or total score calculated by the speech recognition unit 102. If it is within the allowable range, it is determined that the similarity between the advertisement word and the audio data is within the similar allowable range. When the similarity determination unit 130 determines that the similarity between the advertisement word and the voice data is within the similarity allowable range, the display output unit 108 displays and outputs the advertisement word as a voice recognition result of the voice data. In other words, the wider the ranking tolerance range, the higher the possibility that it is determined that the advertising word is similar to the input voice data, and the advertising word is easily displayed and output.

FIG. 9 is a diagram showing an example of the configuration of the advertising word storage unit 120 in the present embodiment.
The advertising word storage unit 120 includes an advertising word column, an advertising fee column, and a rank allowable range column. The advertising word column and the advertising fee column can be the same as the configuration shown in FIG. In the rank allowable range column, for example, a rank allowable range such as “6” or “3” is stored. The allowable rank range can be either an allowable rank range based on the total score or an allowable rank range based on the acoustic score depending on the processing of the similarity determining unit 130.

  Again, for example, suppose that the input voice is “I went to life”. In addition, as a speech recognition result of a word corresponding to “enichi”, it is assumed that “sales day”,..., “NEC”. Here, it is assumed that, when arranged as a candidate for the speech recognition result in descending order of the total score, “Funny Day” is first and “NEC” is third. In this case, the rank of the advertisement word “NEC” shown in FIG. 9 is within the allowable range “6”. For this reason, the similarity determination unit 130 determines that the similarity between the advertising word “NEC” and the voice data is within the allowable range of similarity, and the display output unit 108 indicates “I went to NEC” or the like as the voice recognition result. Output.

  Also in the present embodiment, like the score correction values described in the first embodiment, the allowable range of scores and the allowable range of ranks are within the range of the difference from the input voice data. The upper limit value can be determined in consideration of whether people feel similar. Also in the present embodiment, the allowable range of the score and the allowable range of the ranking may be set so as to be higher as the rank of the advertising fee of the advertising word is higher than the upper limit value determined as described above. it can. Also in this embodiment, the same effect as that of the first embodiment can be obtained.

(Third embodiment)
FIG. 10 is a block diagram showing an example of the configuration of the advertisement display system in the present embodiment.
This embodiment is different from the second embodiment in that the output adjustment unit 104 includes a similar word extraction unit 132 in addition to the configuration shown in FIG. Further, in the present embodiment, the content that the advertising word storage unit 120 stores as a parameter that defines the similar allowable range is also different from that of the second embodiment.

  FIG. 11A is a diagram illustrating an example of the configuration of the advertising word storage unit 120 in the present embodiment. The advertising word storage unit 120 sets the number of similar words similar to the advertising word that can be registered as the parameter that defines the similar allowable range, the higher the advertising fee rank of the advertising word is higher, Is stored in association with. Here, for example, the registerable number of advertisement words “NEC” is set to “20”.

  The similar word extraction unit 132 extracts as many similar words as the adwords that can be registered. The similar word extraction unit 132 uses the acoustic model storage unit 110 and the language model storage unit 112, for example, as the voice recognition unit 102 performs the voice recognition processing, and uses the voice data of the advertising word as input. Recognition processing can be performed. The similar word extraction unit 132 can extract words selected as a result of the speech recognition process as similar words that are similar to the advertisement word in the order of the total score or the acoustic score in the order of the maximum number that can be registered.

  The similar word extracting unit 132 stores the extracted similar words in the advertising word storage unit 120 in association with each advertising word. An example of this is shown in FIG. Here, for example, as a similar word of the advertising word “NEC”, “Family day”, “English character” and the like are stored. In the above procedure, the similar word extraction unit 132 can request the speech recognition unit 102 to perform speech recognition processing and receive the result of the speech recognition processing from the speech recognition unit 102.

  In the present embodiment, the similarity determining unit 130, when an advertisement word or a similar word is cited as the word selected by the speech recognition unit 102 as the speech recognition result of the speech data, the similarity between the advertisement word and the speech data. Is determined to be within the similar allowable range. When the similarity determination unit 130 determines that the similarity between the advertisement word and the voice data is within the similarity allowable range, the display output unit 108 displays and outputs the advertisement word as a voice recognition result of the voice data. In other words, the higher the number of registered similar words, the higher the possibility that it is determined that the advertising word is similar to the input voice data, and the advertising word is easily displayed and output.

  In another example, as a parameter for defining the similarity allowable range, the number of similar words that are similar to the number of advertisement words that are higher as the advertisement charge rank of the advertisement word is higher are associated with the advertisement word and the advertisement word. It can also be registered in advance in the storage unit 120. In this example, the advertising word storage unit 120 has a configuration as shown in FIG.

  In this example, for example, a similar word can be selected by the following procedure. First, using the function of the speech recognition unit 102 or another speech recognition processing system, speech recognition processing is performed on speech data that utters an advertising word, and a total score of a plurality of words that are candidates for speech recognition processing results or Each acoustic score is calculated. Next, a plurality of words whose total score or acoustic score is higher or whose total score or acoustic score is a predetermined value or more are extracted. At this time, it is possible to extract a number of words corresponding to the number of similar words registered in the advertising word storage unit 120 and register them as similar words in the advertising word storage unit 120. Then, the above-mentioned number of words desired by the advertiser can be selected from them and registered in the advertising word storage unit 120 as similar words.

  In this case, the output adjustment unit 104 of the advertisement display system 100 can be configured not to include the similar word extraction unit 132. That is, in this case, the configuration of the advertisement display system 100 can be the same as the configuration shown in FIG.

  The similarity determination unit 130 determines that the similarity between the advertisement word and the voice data is within the allowable range when the advertisement word or the similar word is given as the word selected by the voice recognition unit 102 as the voice recognition result of the voice data. Judge that there is. When the similarity determination unit 130 determines that the similarity between the advertisement word and the voice data is within the similarity allowable range, the display output unit 108 displays and outputs the advertisement word as a voice recognition result of the voice data.

  Also in the present embodiment, like the score correction values described in the first embodiment, the number that can be registered is similar to a person if the difference from the input voice data is within a range. The upper limit value can be determined in consideration of the above. Also in the present embodiment, the number that can be registered can be set to be higher as the rank of the advertising fee of the advertising word is higher within the range of the upper limit determined in this way. Also in this embodiment, the same effect as that of the first embodiment can be obtained.

(Fourth embodiment)
FIG. 12 is a block diagram showing an example of the configuration of the advertisement display system in the present embodiment.
The advertisement display system 100 is different from the configuration described in the first embodiment in that it further includes a mode setting unit 134 in addition to the configuration shown in FIG. In the present embodiment, the output adjustment unit 104 performs normal mode processing in which the voice recognition result selected by the voice recognition unit 102 is output as it is based on the setting of the mode setting unit 134 and the result adjusted by the output adjustment unit 104. Select and execute one of the advertisement mode processes to be output.

  For example, in the procedure described in the first embodiment, if the degree of similarity between the advertising word and the input voice data is within the allowable similarity range, the total score of the result of the voice recognition processing is compared with other words. Even if it is not the best, the advertisement word is displayed and output as a voice recognition result. For this reason, the user who obtains the result of the voice recognition processing with high accuracy is unusable. Therefore, in the present embodiment, the advertisement display system 100 can be configured to be able to select normal mode processing in which the result of normal speech recognition processing is output.

  The mode setting unit 134 can include a setting storage unit that stores which of a normal mode for performing normal voice recognition processing and an advertisement mode is set. When the speech recognition unit 102 performs speech recognition processing, the score adjustment unit 106 determines which one of the advertisement mode and the normal mode is set. When the normal mode is set, the score adjustment unit 106 displays and outputs the result of the speech recognition process of the speech recognition unit 102 as it is from the display output unit 108. On the other hand, when the advertisement mode is set, the score adjustment unit 106 performs the adjustment process as described above so that the advertisement word is easily displayed and output, and the display output unit 108 displays the result after the adjustment process. To do.

FIG. 13 is a flowchart showing a processing procedure of the advertisement display system 100 in the present embodiment.
The processing in steps S101 and S102 can be the same as the procedure described with reference to FIG. 3 in the first embodiment. Thereafter, the score adjustment unit 106 determines which of the advertisement mode and the normal mode is set (step S103). When the advertisement mode is set (YES in step S103), the score adjustment corresponding to the advertisement fee is performed for the advertisement word (step S104). The display output unit 108 outputs the word with the highest score as a speech recognition result based on the score after the score adjustment by the score adjustment unit 106 (step S106).

  On the other hand, in the case where the advertisement mode is not set in step S103 and the normal mode is set (NO in step S103), the score adjustment unit 106 does not perform the adjustment process, and the display output unit 108 displays the voice of the voice recognition unit 102. Based on the score of the recognition process, the word with the highest score is output as a speech recognition result (step S106).

  Note that the advertisement display system 100 described in the second embodiment and the third embodiment also includes the mode setting unit 134 in the same manner, and the normal mode in which the result of the normal voice recognition process is output. The processing can also be selected.

  Although not shown here, the advertisement display system 100 can have a function of identifying a user by, for example, a user ID, and the user can determine whether to use the normal mode or the advertisement mode. it can. For example, for a user who has not registered yet and uses the voice recognition processing function of the voice recognition unit 102 as a trial version, the result of the voice recognition processing in the advertisement mode may be displayed and output. . On the other hand, for a user who has registered as a user and paid a service usage fee, for example, the result of the speech recognition processing in the normal mode may be displayed and output.

  As mentioned above, although embodiment of this invention was described with reference to drawings, these are the illustrations of this invention, Various structures other than the above are also employable.

  As shown in FIG. 15, the advertisement display system 100 can further include a billing processing unit 140 and a billing information storage unit 142. According to the result of the speech recognition process performed by the speech recognition unit 102, the billing processing unit 140 displays the advertisement word as a result of the adjustment process performed by the output adjustment unit 104 even though the advertisement word is not originally displayed and output as a result. That is, the number of times the advertisement word is displayed and output from the display output unit 108 is counted even though the original total score of the advertisement word with respect to the input voice data is lower than the total score of other words. The billing processing unit 140 calculates a charge for each advertisement word based on the number of outputs and the advertisement charge set in the advertisement word storage unit 120.

  The charging processing unit 140 does not charge when the voice data input to the voice recognition unit 102 is an advertisement word from the beginning. For example, the billing processing unit 140 compares the result of the speech recognition process output from the speech recognition unit 102 with the result output from the display output unit 108, and the advertisement word is displayed and output by the adjustment process of the output adjustment unit 104. In this case, the number of outputs can be counted. The billing processing unit 140 can store the number of outputs and the charge in the billing information storage unit 142 for each advertising word. Here, the configuration in which the advertisement display system 100 described in the first embodiment includes the billing processing unit 140 and the billing information storage unit 142 is shown, but the advertisement display in the second to fourth embodiments is shown. The system 100 can also include a charging processing unit 140 and a charging information storage unit 142.

  In the above embodiment, the display output unit 108 has been configured to display and output only one word having the highest total score. However, the display output unit 108 displays the word having the highest total score. It is also possible to output a plurality of items in descending order of the total score so that the user can select them.

  Further, after the advertisement word is displayed and output from the display output unit 108 and the advertisement word is displayed on the display of the user terminal 200, for example, when the user clicks on the advertisement word, the normal voice before the adjustment processing by the output adjustment unit 104 is performed. It is also possible to adopt a configuration in which words as a result of the original speech recognition process that has a high total score as a result of the recognition process are displayed.

  Further, after the advertisement word is displayed and output from the display output unit 108 and the advertisement word is displayed on the display of the user terminal 200, for example, when the user clicks the advertisement word, the site provided by the advertiser of the advertisement word is accessed. It can also be set as the structure which can be performed.

  Each component of the advertisement display system 100 shown in FIG. 1, FIG. 7, FIG. 10, FIG. 12 and FIG. 15 is not a hardware unit configuration but a functional unit block. Each component of the advertisement display system 100 is centered on an arbitrary computer CPU, memory, a program for realizing the components shown in the figure loaded in the memory, a storage unit such as a hard disk for storing the program, and a network connection interface. It is realized by any combination of hardware and software. It will be understood by those skilled in the art that there are various modifications to the implementation method and apparatus.

DESCRIPTION OF SYMBOLS 100 Advertisement display system 102 Speech recognition part 104 Output adjustment part 106 Score adjustment part 108 Display output part 110 Acoustic model memory | storage part 112 Language model memory | storage part 120 Advertising word memory | storage part 130 Similarity judgment part 132 Similar word extraction part 134 Mode setting part 140 Charge Processing unit 142 Accounting information storage unit 150 Network 200 User terminal

Claims (12)

Based on the acoustic model and the language model, the acoustic score of the words that are candidates for the speech recognition result for the input speech data, the language score, and the total score based on the acoustic score and the language score are calculated, and the total score is high. Speech recognition means for selecting words as speech recognition results;
Including an advertising word storage means for storing an advertising word for which an advertising fee is paid as a consideration for facilitating selection as a speech recognition result, and the similarity between the advertising word and the voice data is within a predetermined similarity allowable range Adjusting so that the advertisement word is selected as the voice recognition result of the voice data, and output adjusting means for displaying and outputting the adjusted result as the voice recognition result of the voice data;
Including
The advertisement display system in which the similarity allowable range is set wider as the advertisement charge rank of the advertisement word is higher. The advertisement display system according to claim 1,
The advertisement word storage means stores a correction value of a score that is higher as the advertisement charge rank of the advertisement word is higher as a parameter for defining the similarity allowable range, in association with the advertisement word,
The output adjusting means is
For the advertising word, score adjustment means for performing adjustment to add a correction value of the score to the total score or the language score calculated by the voice recognition means;
Display output means for displaying and outputting a word having a high total score as a voice recognition result of the voice data based on the score adjusted by the score adjusting means;
Advertising display system including. The advertisement display system according to claim 1,
The advertising word storage means stores information indicating an advertising fee of the advertising word in association with the advertising word,
The output adjusting means is
Correction value setting means for setting a correction value of a score that becomes higher as the rank of the advertisement fee of the advertisement word is higher, as a parameter that defines the similar allowable range, based on information indicating the advertisement fee of the advertisement word;
For the advertising word, score adjusting means for performing adjustment to add a correction value of the score to the total score or the language score calculated by the voice recognition means;
Display output means for displaying and outputting a word having a high total score as a voice recognition result of the voice data based on the score adjusted by the score adjusting means;
Advertising display system including. The advertisement display system according to claim 2 or 3,
The score correction value is a language score correction value,
The said score adjustment means is an advertisement display system which performs adjustment which adds the correction value of the said language score to the said language score which the said speech recognition means calculates about the said advertising word. The advertisement display system according to claim 1,
The advertising word storage means stores, as a parameter for defining the similarity allowable range, an allowable range of a score that becomes higher as the advertising fee rank of the advertising word is higher in association with the advertising word,
The output adjusting means is
The difference in score between the acoustic score or the total score of the top word with the highest acoustic score or the total score calculated by the voice recognition means and the acoustic score or the total score of the advertising word is calculated for the voice data. Similarity determination means for determining whether the score is within an allowable range, and determining that the similarity between the advertising word and the audio data is within the allowable range if the score is within the allowable range;
Display output means for displaying and outputting the advertisement word as a voice recognition result of the voice data when the similarity judgment means determines that the similarity between the advertisement word and the voice data is within the similarity allowable range;
Advertising display system including. The advertisement display system according to claim 1,
The advertising word storage means stores, as a parameter defining the similarity allowable range, an allowable range of a rank that becomes a higher value as the advertising fee rank of the advertising word is higher in association with the advertising word,
The output adjusting means is
When the speech data is arranged as a speech recognition result candidate in descending order of the acoustic score or total score calculated by the speech recognition means, it is determined whether or not the rank of the advertising word is within the allowable range, Similarity determination means for determining that the similarity between the advertising word and the audio data is within the similarity allowable range when within the range;
Display output means for displaying and outputting the advertisement word as a voice recognition result of the voice data when the similarity judgment means determines that the similarity between the advertisement word and the voice data is within the similarity allowable range;
Advertising display system including. The advertisement display system according to claim 1,
The advertisement word storage means sets the number of similar words similar to the advertisement word that can be registered as a parameter that defines the similarity allowable range as the advertisement charge rank of the advertisement word is higher. And store it in association with
The output adjusting means is
Similar word extraction means for extracting similar words similar to the advertisement word by the registerable number;
When the advertisement word or the similar word is cited as the word selected as the speech recognition result of the speech data by the speech recognition means, the similarity between the advertisement word and the speech data is within the similarity allowable range. Similarity determination means for determining that there is,
Display output means for displaying and outputting the advertisement word as a voice recognition result of the voice data when the similarity judgment means determines that the similarity between the advertisement word and the voice data is within the similarity allowable range;
Advertising display system including. The advertisement display system according to claim 1,
The advertisement word storage means associates, as a parameter for defining the similarity allowable range, a similar number of similar words that are similar to the advertisement word with a higher value as the advertisement charge rank of the advertisement word is higher. Remember,
The output adjusting means is
When the advertisement word or the similar word is cited as the word selected as the speech recognition result of the speech data by the speech recognition means, the similarity between the advertisement word and the speech data is within an allowable range. Similarity determination means for determining that the similarity is within the allowable range;
Display output means for displaying and outputting the advertisement word as a voice recognition result of the voice data when the similarity judgment means determines that the similarity between the advertisement word and the voice data is within the similarity allowable range;
Advertising display system including. The advertisement display system according to any one of claims 1 to 8,
The output adjustment unit selects, based on the setting, one of a normal mode process for outputting the voice recognition result selected by the voice recognition unit as it is and an advertisement mode process for outputting the result adjusted by the output adjustment unit And display advertising system. The advertisement display system according to any one of claims 1 to 9,
Although the total score calculated by the voice recognition unit is lower than other words, the number of times the advertising word is displayed and output as a voice recognition result as a result of adjustment by the processing adjustment unit is counted. An advertisement display system further including billing processing means for calculating an advertisement fee in response. An advertisement display method using a computer system including an advertisement word storage means for storing an advertisement word for which an advertisement fee is paid as a price for facilitating selection as a speech recognition result,
Based on the acoustic model and the language model, an acoustic score, a language score, and a total score based on the acoustic score and the language score are calculated for a word that is a speech recognition result candidate for the input speech data. A speech recognition step for selecting high words as speech recognition results;
When the similarity between the advertisement word and the voice data is within a predetermined similarity tolerance range, the advertisement word is adjusted to be selected as a voice recognition result of the voice data, and the adjusted result is the voice data. Output adjustment step to display and output as a voice recognition result of
Including
The advertisement display method set so that the similarity allowable range becomes wider as the advertisement charge rank of the advertisement word is higher. Computer
Based on the acoustic model and the language model, an acoustic score, a language score, and a total score based on the acoustic score and the language score are calculated for a word that is a speech recognition result candidate for the input speech data. Speech recognition means for selecting high words as speech recognition results,
Including an advertising word storage means for storing an advertising word for which an advertising fee is paid as a consideration for facilitating selection as a speech recognition result, and the similarity between the advertising word and the voice data is within a predetermined similarity allowable range Adjusting the output so that the advertisement word is selected as the voice recognition result of the voice data, and displaying and outputting the adjusted result as the voice recognition result of the voice data;
A program that functions as
An advertisement display program that is set such that the higher the rank of the advertisement charge of the advertisement word, the wider the similar allowable range.

Images