Information processing method, information processing system, and program

Abstract

To correct a string of characters estimated by speech recognition to a string of characters in appropriate notation.SOLUTION: An information processing system comprises: a speech recognition unit 71 for obtaining a recognition string Z that is a result of speech recognition for speech data; a request transmission unit 721 for transmitting a search request Q including one or more related words other than a target word that is a proper noun included in the recognition string Z to an information retrieval system; a result receiving unit 722 for receiving a search result R in response to the search request Q from the information retrieval system; and a word replacement unit 723 for replacing the target word in the recognition string Z with an alternative word corresponding to the search result.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present disclosure relates to a technique for identifying a character string corresponding to audio data. [Background technology]

[0002] Various speech recognition techniques have been proposed for estimating character strings corresponding to speech data. For example, Patent Document 1 discloses a technique for varying the degree of emphasis of each character in a speech recognition result depending on the confidence level of the recognition result. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 6-131493 Summary of the Invention [Problem to be solved by the invention]

[0004] Incidentally, for proper nouns such as personal names, there are many character strings that are homophones but have different characters. In practice, it is difficult to identify an appropriate character string from multiple character strings that are homophones but have different characters by speech recognition alone. Note that, while character strings with homophones but different characters have been exemplified above for convenience, situations in which an appropriate character string cannot be estimated by speech recognition are not limited to the above examples. In consideration of the above circumstances, one aspect of the present disclosure aims to correct a character string estimated by speech recognition to a character string with an appropriate spelling. [Means for solving the problem]

[0005] In order to solve the above problems, an information processing method according to one aspect of the present disclosure includes obtaining a recognized string that is the result of speech recognition of speech data, transmitting a search request to an information retrieval system that includes one or more related words other than a target word that is a proper noun included in the recognized string, receiving search results in response to the search request from the information retrieval system, and replacing the target word in the recognized string with a substitute word that corresponds to the search result.

[0006] An information processing system according to one aspect of the present disclosure includes a string acquisition unit that acquires a recognized string that is the result of speech recognition of speech data; a request sending unit that sends a search request to an information retrieval system, the search request including one or more related words other than a target word that is a proper noun included in the recognized string; a result receiving unit that receives search results corresponding to the search request from the information retrieval system; and a word replacement unit that replaces the target word in the recognized string with a substitute word that corresponds to the search result.

[0007] A program according to one aspect of the present disclosure causes a computer system to function as a string acquisition unit that acquires a recognized string that is the result of speech recognition of speech data, a request sending unit that sends a search request to an information retrieval system that includes one or more related words other than a target word that is a proper noun included in the recognized string, a result receiving unit that receives search results in response to the search request from the information retrieval system, and a word replacement unit that replaces the target word in the recognized string with an alternative word that corresponds to the search result. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a block diagram illustrating a configuration of an information system according to a first embodiment. [Figure 2] FIG. 2 is a block diagram illustrating a configuration of a terminal device. [Figure 3] FIG. 2 is a schematic diagram of a playback screen. [Figure 4] FIG. 2 is a block diagram illustrating a functional configuration of a terminal device. [Figure 5]FIG. 10 is an explanatory diagram of an operation for correcting a recognized character string. [Figure 6] 10 is a flowchart illustrating a detailed procedure of a control process. [Figure 7] FIG. 10 is an explanatory diagram of an operation for correcting a recognized character string in the second embodiment. [Figure 8] FIG. 10 is a block diagram illustrating the configuration of an information system according to a third embodiment. [Figure 9] FIG. 1 is a block diagram illustrating a configuration of a control system. [Figure 10] FIG. 10 is a block diagram illustrating the configuration of an information system according to a fourth embodiment. [Figure 11] FIG. 10 is a block diagram illustrating an example of the functional configuration of a terminal device according to a fourth embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] A: First embodiment FIG. 1 is a block diagram of an information system 100 in the first embodiment. The information system 100 is a computer system for providing various information to users of terminal devices 30, and includes an information retrieval system 10, a distribution system 20, and a terminal device 30. The terminal device 30 is an information device such as a mobile phone, a smartphone, a tablet terminal, or a personal computer. The terminal device 30 communicates with each of the information retrieval system 10 and the distribution system 20 via a communication network 200 such as the Internet. Although a large number of terminal devices 30 actually exist, the following description will focus on one terminal device 30 for convenience.

[0010] The information retrieval system 10 is a computer system that provides an information retrieval service to a terminal device 30. The information retrieval service is an information service that searches for information in response to a request Q from the terminal device 30 (hereinafter referred to as a "search request"). Specifically, the information retrieval system 10 searches for multiple pieces of search information corresponding to a word (hereinafter referred to as a "search keyword") specified by the search request Q by referring to a search database in which various types of information (hereinafter referred to as "search information"), such as documents or web pages, are registered as search candidates. For example, search information containing words that match or are similar to the search keyword is searched for. The search results R in response to the search request Q (hereinafter referred to as a "search result") are transmitted from the information retrieval system 10 to the terminal device 30.

[0011] The distribution system 20 is a computer system that distributes distribution content C to terminal devices 30. The distribution content C is information created for viewing by a large number of users. Specifically, the distribution content C is digital content that includes video data V representing a video and audio data A representing an audio waveform.

[0012] Terminal device 30 is an information processing system that plays distribution content C. Fig. 2 is a block diagram illustrating the configuration of terminal device 30. Terminal device 30 includes a control device 31, a storage device 32, a communication device 33, a display device 34, and a sound emitting device 35. Terminal device 30 may be realized as a single device, or may be realized as multiple devices configured separately from each other.

[0013] The control device 31 is one or more processors that control each element of the terminal device 30. Specifically, the control device 31 is configured by one or more types of processors, such as a central processing unit (CPU), a graphics processing unit (GPU), a sound processing unit (SPU), a digital signal processor (DSP), a field programmable gate array (FPGA), or an application specific integrated circuit (ASIC).

[0014] The storage device 32 is one or more memories that store programs executed by the control device 31 and various data used by the control device 31. For example, a known recording medium such as a semiconductor recording medium or a magnetic recording medium, or a combination of multiple types of recording media, is used as the storage device 32. Note that, for example, a portable recording medium that is detachable from the terminal device 30, or a recording medium that the control device 31 can access via the communication network 200 (for example, cloud storage) may also be used as the storage device 32.

[0015] The communication device 33 communicates with each of the information retrieval system 10 and the distribution system 20 via the communication network 200. For example, the communication device 33 transmits a search request Q to the information retrieval system 10 and receives a search result R corresponding to the search request Q from the information retrieval system 10. The communication device 33 also receives distribution content C from the distribution system 20. Note that communication between the communication device 33 and the communication network 200 may be either wired or wireless. Furthermore, the communication device 33, which is separate from the terminal device 30, may be connected to the terminal device 30 by wire or wirelessly.

[0016] The display device 34 displays an image under the control of the control device 31. For example, various display panels such as a liquid crystal display panel or an organic EL (Electroluminescence) panel are used as the display device 34. Note that the display device 34, which is separate from the terminal device 30, may be connected to the terminal device 30 by wire or wirelessly.

[0017] The display device 34 of the first embodiment displays the playback screen G of Fig. 3. The playback screen G includes a moving image M and a recognition character string Z. The moving image M is an image represented by the moving image data V of the distribution content C. The recognition character string Z is a character string (i.e., subtitles) representing the spoken content of the audio that constitutes the distribution content C. Therefore, a person with a hearing impairment who has difficulty hearing the audio of the distribution content C can easily view the distribution content C.

[0018] 2 emits sound waves under the control of the control device 31. For example, the sound emitting device 35 reproduces the sound of the distribution content C. The sound emitting device 35 is, for example, a speaker or headphones. Note that the sound emitting device 35, which is separate from the terminal device 30, may be connected to the terminal device 30 by wire or wirelessly.

[0019] Fig. 4 is a block diagram illustrating an example of the functional configuration of the terminal device 30. The control device 31 realizes a plurality of functions (a voice recognition unit 71 and a correction processing unit 72) by executing a program stored in the storage device 32. Fig. 5 is an explanatory diagram regarding the operation of the control device 31.

[0020] The speech recognition unit 71 in FIG. 4 generates a recognized character string Z by performing speech recognition on the audio data A of the distribution content C. The recognized character string Z is the result of speech recognition on the audio data A, and is expressed as a sequence of one or more words. Portions of the recognized character string Z that can be expressed in kanji are expressed in kanji. FIG. 5 shows an example of a recognized character string Z that reads, "I'd like to ask Moriguchi Yuki of ABC Company about their subtitle system."

[0021] Any known technology may be employed for speech recognition by the speech recognition unit 71. For example, speech recognition utilizes an acoustic model such as an HMM (Hidden Markov Model), a language model representing linguistic constraints, and a word dictionary in which a large number of words are registered. In addition to the recognized character string Z, the speech recognition unit 71 also includes a part-of-speech classification and hiragana notation for each word constituting the recognized character string Z. The part-of-speech classification is identification information that specifies the part of speech of each word. For example, proper nouns such as personal names or place names are specified as the part-of-speech classification. The hiragana notation is a string of hiragana characters that represents each word. As can be understood from the above explanation, the speech recognition unit 71 functions as an element (string acquisition unit) that acquires the recognized character string Z.

[0022] It is actually difficult to accurately estimate the kanji spelling of a proper noun through speech recognition. That is, for a word X (hereinafter referred to as a "target word") whose part of speech classification is a proper noun (especially a person's name) among the multiple words in a recognized character string Z, there is a high possibility that a character string that is homophoneically related to the correct character string will be estimated. That is, the kanji spelling of the target word X may be incorrect. Taking the above into consideration, the correction processing unit 72 in FIG. 4 replaces the target word X in the recognized character string Z with a word (hereinafter referred to as a "substitute word") Y that is homophoneically related to the target word X. That is, the target word X in the recognized character string Z is replaced with the substitute word Y that is written in different kanji. In the following description, a person's name is assumed as the target word X. Therefore, the substitute word Y is a person's name that is homophoneically related to the target word X. In FIG. 5, the name "Moriguchi Yuki" is exemplified as the target word X, and the name "Moriguchi Yuki" that is written in different kanji from the target word X is exemplified as the substitute word Y.

[0023] As illustrated in FIG. 4, the correction processing unit 72 includes a request sending unit 721, a result receiving unit 722, and a word replacing unit 723.

[0024] The request sending unit 721 transmits a search request Q from the communication device 33 to the information retrieval system 10. The search request Q is a query requesting a search for search information including a search keyword specified by the search request Q. The search request Q specifies, as search keywords, a target word X included in a recognition character string Z and one or more related words W included in the recognition character string Z. In the search request Q, the target word X is specified in hiragana, as illustrated in FIG. 5 . Therefore, search results R that are independent of the kanji spelling of the target word X can be obtained. The related words W are words other than the target word X among the multiple words that make up the recognition character string Z. Since the recognition character string Z is likely to be a character string describing the target word X, the related words W are words related to the target word X. For example, words that are frequently used to describe the target word X are specified as the related words W. For example, proper nouns such as the name of a place where the person represented by the target word X is located, the name of a person or organization related to the person, or the name of an event in which the person is involved are specified as the related words W. For example, in Figure 5, "ABC Company," to which the person represented by target word X belongs, and "subtitle system" in which the person represented by target word X is involved are exemplified as related words W. Note that, for example, a date and time or a place related to target word X may be specified as related words W. For example, the date and time of an event in which the person represented by target word X will participate, or the place where the event will be held, are exemplified as related words W.

[0025] The information retrieval system 10 searches a search database for multiple pieces of search information corresponding to a search request Q received from a terminal device 30. Specifically, the search database is searched for multiple pieces of search information that include the search keywords (target word X and related word W) specified by the search request Q. The information retrieval system 10 transmits the search result R to the terminal device 30 that made the request.

[0026] The search result R includes multiple pieces of search information searched from a search database. A search ranking is set for the multiple pieces of search information included in the search result R. For example, the multiple pieces of search information are arranged in the order of their search ranking. The search ranking is the order of priority as a search result. For example, the search ranking of each piece of search information is set according to multiple factors such as the relevance to the search keyword, the frequency with which the search information has been viewed or searched, or the date and time when the search information was updated.

[0027] 4 receives the search result R corresponding to the search request Q from the information retrieval system 10. Specifically, the result receiving unit 722 receives the search result R transmitted by the information retrieval system 10 via the communication device 33.

[0028] The word replacement unit 723 replaces the target word X in the recognition character string Z with a substitute word Y corresponding to the search result R. Specifically, the word replacement unit 723 replaces the target word X in the recognition character string Z with a substitute word Y included in the search information that has the highest search ranking among the multiple pieces of search information included in the search result R. For example, the word replacement unit 723 identifies a substitute word Y corresponding to the target word X from the search information that has the highest search ranking, and replaces the target word X with the substitute word Y. For example, a word that is similar to or matches the target word X among the character strings included in the search information is identified as the substitute word Y. As can be understood from the above explanation, the target word X is replaced with a substitute word Y that should be prioritized from multiple perspectives, such as relevance to the search keyword, how often the search information has been viewed, or the date and time the search information was updated. Specifically, the word replacement unit 723 replaces the target word X with the substitute word Y for the recognition character string Z on the playback screen G displayed on the display device 34.

[0029] 6 is a flowchart of a process (hereinafter referred to as "control process") executed by the control device 31. For example, the control process is executed for each recognition character string Z corresponding to the voice data A. In other words, if the speech content of the voice data A consists of multiple sentences, the control process is repeated for each recognition character string Z (each sentence). However, the division of the recognition character string Z is not limited to one sentence. For example, the control process may be executed for each multiple sentences of the recognition character string Z. The control process is executed in real time in parallel with the distribution of the distribution content C by the distribution system 20.

[0030] When the control process starts, the control device 31 (voice recognition unit 71) identifies a recognized character string Z by voice recognition of the voice data A (S1). The control device 31 (voice recognition unit 71) displays the recognized character string Z on the display device 34 (S2). Specifically, the control device 31 displays the target word X and character strings other than the target word X in the recognized character string Z in different modes. FIG. 5 illustrates a state in which the target word X is highlighted by underlining. Note that the display mode of the target word X is not limited to the above example. For example, various modes such as the display color, type (font), size, decoration (e.g., shading) of the characters of the target word X may differ from those of the character strings other than the target word X.

[0031] The control device 31 (request sending unit 721) sends a search request Q including a target word X and a related word W to the information retrieval system 10 from the communication device 33 (S3). Then, the control device 31 (result receiving unit 722) receives the search result R sent from the information retrieval system 10 by the communication device 33 (S4).

[0032] The control device 31 (word replacement unit 723) replaces the target word X in the recognized character string Z with the substitute word Y corresponding to the search result R (S5). That is, the target word X in the recognized character string Z is replaced with the substitute word Y included in the search information with the highest search ranking. Specifically, as illustrated in FIG. 5, the control device 31 replaces the target word X displayed on the display device 34 with the substitute word Y. Note that if the substitute word Y matches the target word X, replacement of the target word X is not executed.

[0033] As described above, in the first embodiment, a search request Q including a target word X, which is a proper noun in a recognition character string Z, and one or more related words W is transmitted to the information retrieval system 10, and the target word X of the recognition character string Z is replaced with a substitute word Y corresponding to the search result R. Therefore, it is possible to replace a target word X that is erroneously estimated in speech recognition by the speech recognition unit 71 with an appropriate substitute word Y. In other words, it is possible to correct the recognition character string Z to an appropriate notation character string. Furthermore, because the search request Q includes a related word W other than the target word X, there is an advantage that there is a higher possibility of finding an appropriate substitute word Y to replace the target word X, compared to, for example, a mode in which a search request Q including only the target word X is transmitted to the information retrieval system 10.

[0034] Incidentally, among proper nouns, there is a particularly high possibility that a character string having a homophone but different characters will be estimated by speech recognition when it is used as an appropriate character string. In the first embodiment, a character string having a homophone but different characters with respect to a target word X is substituted for the target word X as a substitute word Y. Therefore, it is possible to replace a target word X that has been erroneously estimated as a homophone but different characters in speech recognition with an appropriate substitute word Y.

[0035] Furthermore, in the first embodiment, a recognition character string Z including a target word X is provisionally displayed on the display device 34, and when the search result R is acquired, the target word X displayed on the display device 34 is replaced with the substitute word Y. Therefore, compared to a configuration in which, for example, when the search result R is acquired, the display of the recognition character string Z in which the target word X is replaced with the substitute word Y begins, the user can quickly understand the recognition character string Z. Note that the display of the recognition character string Z including the target word X may be omitted. In other words, only the recognition character string Z in which the target word X is replaced with the substitute word Y may be displayed on the display device 34.

[0036] In the first embodiment, the target word X is replaced with the substitute word Y corresponding to the search information with the highest search ranking among a plurality of pieces of search information searched in response to the search request Q. Therefore, a word with a high priority from various viewpoints such as past search trends can be preferentially applied as the substitute word Y.

[0037] B: Second embodiment A second embodiment will be described. Note that, for elements in the following exemplary aspects that have the same functions as those in the first embodiment, the same reference numerals as those in the first embodiment will be used, and detailed descriptions of each will be omitted as appropriate.

[0038] 7 is an explanatory diagram regarding the operation of the control device 31 in the second embodiment. The control device 31 in the second embodiment executes a program stored in the storage device 32 to realize the same elements (a voice recognition unit 71 and a correction processing unit 72) as those in the first embodiment.

[0039] The search request Q transmitted by the request transmitting unit 721 in the first embodiment includes a target word X and one or more related words W. On the other hand, the search request Q transmitted by the request transmitting unit 721 in the second embodiment specifies one or more related words W as search keywords, similar to those in the first embodiment, but does not include the target word X. In other words, the request transmitting unit 721 transmits a search request Q that does not include the target word X from the communication device 33 to the information retrieval system 10. The configuration and operation other than the content of the search request Q are the same as those in the first embodiment.

[0040] The information retrieval system 10 searches the search database for multiple pieces of search information corresponding to the search request Q received from the terminal device 30. Specifically, the search database is searched for multiple pieces of search information including one or more related words W specified as search keywords by the search request Q. The information retrieval system 10 transmits the search result R to the terminal device 30 that issued the request. As in the first embodiment, the result receiving unit 722 receives the search result R transmitted by the information retrieval system 10 via the communication device 33.

[0041] Search information that includes one or more related words W is likely to be related to the target word X. Search information with a high search ranking is particularly likely to include the target word X. The word replacement unit 723 identifies a substitute word Y that corresponds to the target word X from the search information with the highest search ranking, and replaces the target word X with the substitute word Y. For example, a word that is similar to or matches the target word X among the character strings included in the search information is identified as the substitute word Y. The operation (S2) of the word replacement unit 723 to display the recognized character string Z including the target word X on the display device 34, and the operation (S5) to replace the target word X with the substitute word Y are the same as those in the first embodiment.

[0042] The second embodiment also achieves the same effects as the first embodiment. Furthermore, in the second embodiment, a search request Q that does not include the target word X itself is transmitted to the information retrieval system 10, so there is a high possibility that a word other than the target word X will be searched for as the substitute word Y. For example, a word that is homophoneically related to the target word X and is often used in association with the related word W is searched for as the substitute word Y. In other words, the second embodiment has the advantage that there is a high possibility that an appropriate substitute word Y to replace the target word X will be searched for.

[0043] On the other hand, in the first embodiment, since the search request Q includes the target word X, it is easy to retrieve search information that includes the target word X. In other words, the first embodiment has the advantage of reducing the possibility that search information that is less relevant to the target word X will be retrieved.

[0044] C: Third embodiment Fig. 8 is a block diagram of an information system 100 according to the third embodiment. As illustrated in Fig. 8, the information system 100 according to the third embodiment includes a control system 40 in addition to the same elements as those in the first embodiment. In the first embodiment, the terminal device 30 transmits a search request Q (S3), receives a search result R (S4), and corrects the recognized character string Z (S5). The control system 40 according to the third embodiment is a computer system that executes the processes described above.

[0045] 9 is a block diagram illustrating an example of the configuration of the control system 40. The control system 40 includes a control device 41, a storage device 42, and a communication device 43. The control system 40 may be realized as a single device, or may be realized as multiple devices configured separately from each other.

[0046] The control device 41 is one or more processors that control each element of the control system 40. Specifically, the control device 41 is configured by one or more types of processors, such as a CPU, a GPU, an SPU, a DSP, an FPGA, or an ASIC.

[0047] The storage device 42 is one or more memories that store programs executed by the control device 41 and various data used by the control device 41. For example, a well-known recording medium such as a semiconductor recording medium or a magnetic recording medium, or a combination of multiple types of recording media, is used as the storage device 42. Note that, for example, a portable recording medium that is detachable from the control system 40, or a recording medium that the control device 41 can access via the communication network 200 (for example, cloud storage) may also be used as the storage device 42.

[0048] The communication device 43 communicates with each of the information retrieval system 10 and the distribution system 20 via the communication network 200. For example, the communication device 43 transmits a search request Q to the information retrieval system 10 and receives a search result R corresponding to the search request Q from the information retrieval system 10. The communication device 43 also receives distribution content C from the distribution system 20. Note that communication between the communication device 43 and the communication network 200 may be either wired or wireless. Furthermore, the communication device 43, which is separate from the control system 40, may be connected to the control system 40 by wire or wirelessly.

[0049] The control device 41 of the third embodiment executes a program stored in the storage device 42 to realize the same functions (speech recognition unit 71 and correction processing unit 72) as those of the first embodiment illustrated in Fig. 4. As in the first embodiment, the speech recognition unit 71 generates a recognized character string Z by speech recognition of speech data A of distribution content C. The request sending unit 721 sends a search request Q from the communication device 43 to the information retrieval system 10, and the result receiving unit 722 receives the search result R from the information retrieval system 10 via the communication device 43.

[0050] The word replacement unit 723 replaces the target word X in the recognition character string Z with a substitute word Y corresponding to the search result R. The word replacement unit 723 transmits the replaced recognition character string Z and the distribution content C from the communication device 43 to the terminal device 30. The terminal device 30 plays back the distribution content C and the recognition character string Z in the same manner as in the first embodiment. For example, a playback screen G including a video M of the distribution content C and the recognition character string Z in which the target word X has been replaced with the substitute word Y is displayed on the display device 34 of the terminal device 30. Note that, as in the first embodiment, the recognition character string Z including the target word X may be provisionally displayed on the display device 34, and after the search result R is generated, the displayed target word X may be replaced with the substitute word Y. For example, the word replacement unit 723 may display the recognition character string Z including the target word X on the terminal device 30, and after the search result R is obtained, transmit an instruction to replace the target word X with the substitute word Y from the communication device 43 to the terminal device 30.

[0051] The third embodiment also achieves the same effects as the first embodiment. The configuration of the second embodiment, in which the search request Q does not include the target word X, is similarly applied to the third embodiment. The terminal device 30 of the first embodiment and the control system 40 of the third embodiment are collectively expressed as an information processing system that replaces the target word X of the recognition character string Z with the substitute word Y. The presence or absence of the target word X in the search request Q is not an issue in the third embodiment.

[0052] D: Fourth embodiment 10 is a block diagram of an information system 100 according to the fourth embodiment. The information system 100 according to the fourth embodiment includes a voice recognition system 50 in addition to the same elements as those of the first embodiment (information retrieval system 10, distribution system 20, and terminal device 30). The voice recognition system 50 generates a recognized character string Z by voice recognition of voice data A of distribution content C, similar to the voice recognition unit 71 according to the first embodiment.

[0053] 11 is a block diagram illustrating the functional configuration of a terminal device 30 according to the fourth embodiment. The control device 31 according to the fourth embodiment functions as a recognition request unit 73 instead of the voice recognition unit 71 according to the first embodiment. The recognition request unit 73 requests the voice recognition system 50 to perform voice recognition on the voice data A of the distribution content C. Specifically, the recognition request unit 73 transmits a recognition request including the voice data A from the communication device 33 to the voice recognition system 50.

[0054] The voice recognition system 50 generates a recognized character string Z by voice recognition of the voice data A received from the terminal device 30. The voice recognition method is the same as in the first embodiment. The voice recognition system 50 transmits the recognized character string Z identified by the voice recognition to the requesting terminal device 30. The recognition request unit 73 receives the recognized character string Z transmitted from the voice recognition system 50 by the communication device 33. The control processing (S2 to S5) using the recognized character string Z is the same as in the first embodiment.

[0055] The fourth embodiment also achieves the same effects as the first embodiment. In the fourth embodiment, the terminal device 30 does not need to perform voice recognition, which has the advantage of reducing the processing load on the control device 31. As can be understood from the above explanation, the voice recognition unit 71 of the first embodiment and the recognition request unit 73 of the fourth embodiment are collectively expressed as an element (character string acquisition unit) that acquires the recognized character string Z, which is the result of voice recognition of the voice data A. In other words, acquisition of the recognized character string Z includes both the operation of the voice recognition unit 71 of the first embodiment generating the recognized character string Z through voice recognition of the voice data A and the operation of the recognition request unit 73 of the fourth embodiment receiving the recognized character string Z from the voice recognition system 50.

[0056] The configuration of the second embodiment, in which the search request Q does not include the target word X, is also applied to the fourth embodiment. The configuration of the third embodiment, in which the control system 40 transmits the search request Q (S3), receives the search result R (S4), and corrects the recognized character string Z (S5), is also applied to the fourth embodiment. That is, the control system 40 may request the voice recognition system 50 to recognize the voice data A. The recognized character string Z generated by the terminal device 30 (voice recognition unit 71) through voice recognition may be received by the control device 41 (recognition request unit 73) of the control system 40 from the terminal device 30 via the communication device 43. The recognized character string Z received by the terminal device 30 from the voice recognition system 50 may be received by the control device 41 of the control system 40 from the terminal device 30 via the communication device 43.

[0057] E: Modified Example Specific modified embodiments that can be added to each of the above-described embodiments are exemplified below. Multiple embodiments arbitrarily selected from the above-described embodiments and the modified embodiments exemplified below may be combined as appropriate within the scope of not mutually contradicting each other.

[0058] (1) In each of the above-described embodiments, an operation (hereinafter referred to as a “search operation”) was performed for each target word X of the recognition string Z, including sending a search request Q (S3) and receiving search results R (S4). However, if the target word X is used repeatedly in the distribution content C, it is not necessary to perform a search operation for all target words X.

[0059] For example, when a substitute word Y is identified for a specific target word X, the control device 31 (word replacement unit 723) may store the substitute word Y in the storage device 32, and replace target words X that subsequently appear in the recognized character string Z with the substitute word Y stored in the storage device 32. In other words, the search operation may be omitted for the second and subsequent target words X.

[0060] The frequency with which the alternative word Y should be identified also depends on the type of broadcast content C. For example, for broadcast content C that relates to one theme throughout, the target word X may be replaced with a common alternative word Y throughout the broadcast content C. On the other hand, for broadcast content C that includes many sections related to different themes (e.g., a news program), the alternative word Y may need to be changed within the broadcast content C. For example, in a situation where multiple words that have homophones but different characters are used sequentially, the alternative words Y need to be identified sequentially.

[0061] Considering the above circumstances, an example is shown in which the control device 31 selects one of a plurality of operation modes including a first operation mode in which a search operation (S3, S4) is performed each time a target word X occurs, and a second operation mode in which a searched substitute word Y stored in the storage device 32 is used to replace the target word X. In the first operation mode, each target word X can be replaced with an appropriate substitute word Y even in a situation in which multiple words with the same pronunciation but different characters are used alternately. On the other hand, in the second operation mode, the search operation is omitted for the second and subsequent target words X. This reduces the processing load on the control device 31 (word replacement unit 723).

[0062] The operation mode is selected, for example, according to the type of distribution content C. For example, if distribution content C falls into a first type that tends to include multiple sections with different themes, the control device 31 selects the first operation mode. On the other hand, if distribution content C falls into a second type that tends to have a constant theme, the control device 31 selects the second operation mode. According to the above embodiment, it is possible to achieve both the advantages of the first operation mode and the advantages of the second operation mode.

[0063] In the above description, it is assumed that the terminal device 30 executes the search operation, but the same configuration is also applied to a third embodiment in which the control system 40 executes the search operation.

[0064] (2) In the above-described embodiments, the search result R includes multiple pieces of search information. However, the search result R may include one piece of search information. For example, the information retrieval system 10 transmits the search result R including the single piece of search information that has the highest search ranking among the multiple pieces of search information retrieved in response to the search request Q. The word replacement unit 723 replaces the target word X of the recognition character string Z with the substitute word Y included in the search information of the search result R.

[0065] (3) In each of the above-described forms, a person's name is exemplified as the target word X, but the target word X is not limited to these examples. For example, a proper noun such as a group name, a place name, or a country name may be specified as the target word X.

[0066] (4) In the above-described embodiments, the substitute word Y included in the search information with the highest search ranking among the plurality of pieces of search information included in the search result R is used to replace the target word X. However, the method of selecting the search information from which the substitute word Y is to be extracted from the plurality of pieces of search information is not limited to the above example of using the search ranking. In an embodiment in which the most recent date and time at which each piece of search information was generated or updated (hereinafter referred to as the "update date and time") is included in each piece of search information in the search result R, the substitute word Y may be extracted from the search information with the update date and time closest to the current time. Furthermore, for example, the substitute word Y may be extracted from the search information with the highest number of searches among the plurality of pieces of search information. The search information may be selected by comprehensively taking into account a plurality of factors such as the search ranking, the update date and time, and the number of searches.

[0067] (5) In the above-described embodiments, the information retrieval system 10 provides an information retrieval service. However, the information service provided by the information retrieval system 10 is not limited to the information retrieval service. For example, a computer system that manages a social networking service (SNS) where users post and view desired information may be used as the information retrieval service in the above-described embodiments. Upon receiving the search request Q, the information retrieval system 10 transmits, as search result R, posted information that includes a word (so-called trending word or rapidly rising word) that has been posted or viewed many times in a predetermined period up to the current time, from among a large number of pieces of search information that include the target word X.

[0068] (6) In the above-described embodiments, the recognition character string Z before and after replacement of the target word X is displayed on the display device 34, but the method of outputting the recognition character string Z is not limited to the above examples. For example, the control device 31 (word replacement unit 723) may replace the target word X of the recognition character string Z stored in the storage device 32 with a substitute word Y, and transmit the replaced recognition character string Z from the communication device 33 to another information device. In other words, the display of the recognition character string Z (before or after replacement) may be omitted.

[0069] (7) In the above-described embodiments, information such as documents or web pages has been exemplified as search information, but the content of the search information is not limited to these examples. For example, the information providing system may transmit search results R including location information indicating the location of various pieces of information. The location information may be, for example, a uniform resource locator (URL). The word replacement unit 723 obtains information such as documents by accessing the system indicated by the URL of the search information, and extracts a substitute word Y corresponding to the target word X from the information.

[0070] (8) In the above-described embodiments, the voice data A to be subjected to voice recognition is distributed from the distribution system 20. However, the source of the voice data A is not limited to the distribution system 20. For example, the terminal device 30 may acquire the voice data A from a sound collection device (microphone) that generates the voice data A by collecting ambient sounds. The terminal device 40 may acquire the voice data A from various types of audio equipment connected to the terminal device 30 by wire or wirelessly. The terminal device 30 may receive the voice data A transmitted from another communication device via the communication network 200. Furthermore, the voice data A does not need to be data that constitutes the distribution content C together with the video data V. For example, the voice data A may be distributed independently.

[0071] (9) As described above, the functions of the terminal device 30 according to each of the above-described embodiments are realized by the cooperation of one or more processors constituting the control device 31 and the program stored in the storage device 32. Also, the functions of the control system 40 according to the third embodiment are realized by the cooperation of one or more processors constituting the control device 41 and the program stored in the storage device 42.

[0072] The programs exemplified above can be provided in a form stored on a computer-readable recording medium and installed on a computer. The recording medium is, for example, a non-transitory recording medium, such as an optical recording medium (optical disk) such as a CD-ROM, but also includes any known type of recording medium, such as a semiconductor recording medium or a magnetic recording medium. Note that a non-transitory recording medium includes any recording medium other than a transitory, propagating signal, and does not exclude volatile recording media. In addition, in a configuration in which a distribution device distributes a program via a communication network, the recording medium that stores the program in the distribution device corresponds to the non-transitory recording medium described above.

[0073] F: Notes From the above-described exemplary embodiments, the following configurations can be understood, for example.

[0074] An information processing method according to one aspect (aspect 1) of the present disclosure includes obtaining a recognized string that is the result of speech recognition of speech data, transmitting a search request to an information retrieval system that includes one or more related words other than a target word that is a proper noun included in the recognized string, receiving search results corresponding to the search request from the information retrieval system, and replacing the target word in the recognized string with an alternative word that corresponds to the search result.

[0075] According to the above aspects, a search request including one or more related words other than a target word, which is a proper noun in a recognition string, is transmitted to an information retrieval system, and the target word in the recognition string is replaced with a substitute word corresponding to a search result in response to the search request. Therefore, it is possible to replace a target word that is erroneously estimated in speech recognition with an appropriate substitute word. In other words, it is possible to correct the recognition string to a character string with an appropriate spelling. Furthermore, since the search request includes related words other than the target word, there is an advantage that, compared to a form in which a search request including only the target word is transmitted to an information retrieval system, there is a higher possibility that an appropriate substitute word to replace the target word will be found.

[0076] The speech recognition can be realized by the information processing system itself that realizes the information processing method, and by a speech recognition system with which the information processing system can communicate. In other words, "obtaining a recognition character string" encompasses both an operation in which the information processing system itself estimates a recognition character string through speech recognition, and an operation in which the information processing system receives a recognition character string estimated by an external speech recognition system.

[0077] A "target word" is a proper noun included in the recognition character string. For example, the names (pronunciations) of various things such as people or place names are examples of "target words." A target word may consist of a single word or multiple words connected together.

[0078] "Related words" are words other than the target word in the recognition string. For example, words related to the thing represented by the target word are exemplified as "related words." For example, a proper noun (e.g., a name) that represents the location of the thing represented by the target word is an example of a "related word."

[0079] Each word that makes up a recognition character string can correspond to either a “target word” or a “related word.” For example, if a recognition character string includes a first word and a second word, when the first word is the target word, the second word corresponds to a related word, and when the second word is the target word, the first word corresponds to a related word.

[0080] An information retrieval system is a computer system that provides information services. An information service is a service that searches for and provides information in response to a search request from an external device. Examples of "information retrieval systems" include computer systems that provide various information services, such as an information retrieval service that searches for various information, such as websites or files that contain words specified in a search request, or a social media service (SNS) that provides information posted by each user to a large number of users.

[0081] The "search results" are provided to the information processing system in the form of, for example, the URL of a website containing the words specified in the search request, or text data containing the words. The information search system prioritizes providing search results based on information that is of interest to many users. For example, information that has been searched for frequently in an information search service, or information that has been posted or viewed frequently in a social networking service, is prioritized as search results.

[0082] A "substitute word" is a substitute word that should replace the target word. For example, the substitute word is a word that is homonymous with the target word. The substitute word corresponding to the search result is, for example, a word that can be identified from the search result. Specifically, when the search result is provided as a website URL, words contained in the website are exemplified as substitute words. Furthermore, when the search result is provided as data such as text data, words contained in the text are exemplified as substitute words. The target word in the recognition string is substituted with the substitute word. However, there may be cases where the target word is not substituted with the substitute word.

[0083] In a specific example (Aspect 2) of Aspect 1, the search request includes the target word. In the above aspect, a search request including both the target word and related words is sent to the information retrieval system, which increases the likelihood that the information retrieval system will find an appropriate substitute word to replace the target word.

[0084] In a specific example (Aspect 3) of Aspect 1, the search request does not include the target word. In the above aspect, a search request that does not include the target word is sent to the information retrieval system, so there is a high possibility that words other than the target word will be searched for as substitute words. For example, words that are homophones with different characters to the target word and that are often used in conjunction with related words will be searched for as substitute words.

[0085] In a specific example (Aspect 4) of any one of Aspects 1 to 3, the target word is a person's name, and the substitute word is a person's name that has a homophone but different characters with the target word. Among proper nouns, there is a particularly high possibility that a character string that has a homophone but different characters will be estimated by speech recognition compared to an appropriate character string for a person's name. By configuring a person's name that has a homophone but different characters with the target word as a substitute word, it is possible to replace a target word that has been erroneously estimated with a homophone but different characters in speech recognition with an appropriate substitute word. Note that "homophone but different characters" refers to a relationship in which the pronunciation (name) is the same but the spelling (especially kanji) is different.

[0086] A specific example (Aspect 5) of any of Aspects 1 to 4 further includes displaying the target word and the character string other than the target word in the recognition character string in different modes on a display device, and replacing with a substitute word includes replacing the target word displayed on the display device with the substitute word. In the above aspects, a recognition character string including the target word is provisionally displayed on a display device, and when search results are obtained, the target word displayed on the display device is replaced with a substitute word corresponding to the search results. Therefore, compared to a configuration in which, for example, when search results are obtained, display of a recognition character string in which the target word has been replaced with a substitute word is started, a user can quickly understand the recognition character string.

[0087] The "appearance" of an image displayed on a display device refers to the characteristics of the image that can be visually distinguished by an observer. For example, the displayed color, pattern (design), size, or shape of a character string or background is included in the concept of "appearance." The "displayed color" is defined by hue (tone), saturation, or brightness (gradation).

[0088] In a specific example (Aspect 6) of any of Aspects 1 to 5, obtaining the search results includes obtaining search results including a plurality of pieces of search information that have been searched by the information retrieval system and have been set in search rankings, and replacing the target word includes replacing the target word with the substitute word corresponding to the search information that has the highest search ranking among the plurality of pieces of search information. In the above aspects, the target word is replaced with the substitute word corresponding to the search information that has the highest search ranking among the plurality of pieces of search information searched by the information retrieval system. Therefore, words with high priority from various perspectives such as past search trends can be preferentially applied as substitute words.

[0089] A "search ranking" is a ranking determined based on predetermined conditions for multiple search results. Specifically, information that attracts the attention of many users is ranked high in the search ranking. For example, information that is searched frequently in an information search service, or information that is posted or viewed frequently in a social media service, is ranked high in the search ranking.

[0090] An information processing system according to one aspect (aspect 7) of the present disclosure includes a string acquisition unit that acquires a recognized string that is the result of speech recognition of speech data, a request sending unit that sends a search request to an information retrieval system that includes one or more related words other than a target word that is a proper noun included in the recognized string, a result receiving unit that receives search results in response to the search request from the information retrieval system, and a word replacement unit that replaces the target word in the recognized string with an alternative word that corresponds to the search result.

[0091] A program according to one aspect (aspect 8) of the present disclosure causes a computer system to function as a string acquisition unit that acquires a recognized string that is the result of speech recognition of speech data, a request sending unit that sends a search request to an information retrieval system that includes one or more related words other than a target word that is a proper noun included in the recognized string, a result receiving unit that receives search results in response to the search request from the information retrieval system, and a word replacement unit that replaces the target word in the recognized string with an alternative word that corresponds to the search result. [Explanation of symbols]

[0092] 100...information system, 200...communication network, 10...information search system, 20...distribution system, 30...terminal device, 31...control device, 32...storage device, 33...communication device, 34...display device, 35...sound emission device, 40...control system, 41...control device, 42...storage device, 43...communication device, 50...speech recognition system, 71...speech recognition unit, 72...correction processing unit, 721...request sending unit, 722...result receiving unit, 723...word replacement unit, 73...recognition request unit.

Claims

1. obtaining a recognized character string that is a result of speech recognition of the speech data; sending a search request to an information retrieval system, the search request including one or more related words other than the target word, which is a proper noun included in the recognized character string; receiving, from the information retrieval system, a search result including a plurality of pieces of search information that have been searched by the information retrieval system in response to the search request and that have been set in search order; and replacing the target word in the recognized character string with a substitute word corresponding to the search information having the highest search ranking among the plurality of search information in the search results. An information processing method implemented by a computer system.

2. The search request includes the target word. The information processing method according to claim 1.

3. The search request does not include the target word. The information processing method according to claim 1.

4. the target word is a person's name, The alternative word is a person's name that has a homophone but different characters from the target word.

4. The information processing method according to claim 1.

5. The method further includes displaying the target word and a character string other than the target word in the recognized character string in different modes on a display device, The step of replacing the target word with the substitute word includes replacing the target word displayed on the display device with the substitute word.

4. The information processing method according to claim 1.

6. a character string acquisition unit that acquires a recognized character string that is a result of speech recognition of the speech data; a request sending unit that sends a search request to an information search system, the search request including one or more related words other than the target word, which is a proper noun included in the recognized character string; a result receiving unit that receives, from the information retrieval system, a search result including a plurality of pieces of search information that have been searched by the information retrieval system in response to the search request and whose search rankings have been set; a word replacement unit that replaces the target word in the recognized character string with a substitute word corresponding to the search information that has the highest search ranking among the plurality of search information in the search result; An information processing system comprising:

7. a character string acquisition unit that acquires a recognized character string that is a result of speech recognition of the speech data; a request sending unit that sends a search request to an information retrieval system, the search request including one or more related words other than the target word, which is a proper noun included in the recognized character string; a result receiving unit that receives, from the information retrieval system, a search result including a plurality of pieces of search information that have been searched by the information retrieval system in response to the search request and that have been set in search order; and a word replacement unit that replaces the target word in the recognized character string with a substitute word corresponding to the search information that has the highest search ranking among the plurality of search information in the search result; A program that makes a computer system function as a

Images