JP2022059732A - Information processing device, control method, and program - Google Patents

Abstract

To provide a technique for confirming and correcting a result of voice recognition in a short time.SOLUTION: The present invention is an information processing device that acquires recognition data as a voice recognition result of voice data and text data corresponding to the voice data, including: specifying means for specifying a portion of attention in the recognition data or the text data; and display control means for, when displaying a result of comparing the recognition data and the text data, identifying and displaying the specified portion.SELECTED DRAWING: Figure 3

Description

The present invention relates to a technique for modifying a recognition result of a speech recognition engine.

As a method of assisting a Deaf person to obtain information in the same way as a listener, there is a method of displaying the utterance content of the speaker as a text. For example, there is also a method of automatically converting an utterance into a character string by subtitle broadcasting on TV, writing a summary at a lecture, or voice recognition, and displaying it on a PC screen or the like.

However, there is a considerable time lag between when they are actually spoken and when Deaf people can read them as letters. In teletext on television, subtitles may be displayed 10 to 15 seconds after the utterance.

For example, assuming a "weather corner", the explanation about "tomorrow's forecast" is over, and after the scene changes to "forecast for the next week", the utterance of the weather forecaster in the previous "forecast tomorrow" is subtitled. In order to understand the contents of the subtitles, it is necessary to memorize the previous figure.

Also, after the video of the news is changed to that of the next incident, subtitles related to the video of the previous incident may be displayed, which is very difficult to understand.

In order to solve this problem, a method of providing subtitles at a timing as close to real time as possible is being studied.

In the technique described in Patent Document 1, a manuscript such as a news program is registered in a text correction device in advance, and the utterance of an announcer or the like is compared with the recognition result obtained by voice recognition. Specifically, the following method is used.

First, assuming that the recognition result of the N morpheme contains some misrecognition, the position of which N morpheme of the manuscript corresponds to is specified. Furthermore, if there is a morpheme that is misrecognized, it is presented by replacing it with the morpheme in the manuscript.

Furthermore, as a second method, it is not determined that the morpheme of the recognition result is incorrect, and the recognition result is compared with the reading of the morpheme of the manuscript. Do not correct the morpheme that was created as correct.

Japanese Unexamined Patent Publication No. 2012-128188

In the technique of Patent Document 1, when the similarity between the morpheme sequence of the recognition result and the morpheme sequence of the previously prepared manuscript is the highest, and the similarity exceeds the threshold value specified in advance, the two are compared and the manuscript is compared. It is determined whether or not to correct (such as step S16 in FIG. 6) or not (when “NO” is obtained in step S28 in FIG. 10).

In other words, it is premised that "the morpheme sequence of the recognition result is in the manuscript" to determine whether to correct or not, and if the speaker makes a statement that is not in the manuscript, the similarity is the threshold value. Since it is determined that the above value is not exceeded (“NO” in step S15 in FIG. 6 and “NO” in step S25 in FIG. 10), the output is performed as it is. Furthermore, the similarity between the recognition result and the manuscript is calculated by a mechanical method comparing morpheme sequences. And all the corrections based on the result are automatic based on the manuscript.

However, confirmation by a human proofreader is still important. In order to emphasize real-time performance, humans do not check and correct all the character strings that are candidates for subtitles, but only the important parts are made to correspond to the proofreader. Depending on the purpose of the subtitles, the points of confirmation that should be limited differ depending on the content of the manuscript, such as when even subtle nuances should not be mistaken depending on the content, or when there are relatively few problems if the numerical values etc. are accurately conveyed. ..

In the case of Japanese, the nuances may change significantly depending on the attached words (for example, particles, auxiliary verbs, interjections, etc. added to the end of the sentence), and it may be better to manually check such morphemes.

An object of the present invention is to provide a technique for confirming and correcting a voice recognition result in a short time.

The present invention is an information processing device that acquires recognition data that is a voice recognition result of voice data and text data corresponding to the voice data, and specifies a portion of interest in the recognition data or the text data. It is characterized by comprising a specific means for identifying and displaying the specified portion when displaying the result of comparing the recognition data and the text data.

INDUSTRIAL APPLICABILITY According to the present invention, it is possible to provide a technique for confirming and correcting the result of voice recognition in a short time.

It is a figure which shows an example of the system configuration which concerns on embodiment of this invention. It is a block diagram which shows an example of the hardware composition of the voice recognition server and the information processing terminal which concerns on embodiment of this invention. It is a figure which shows an example of the functional structure which concerns on embodiment of this invention. This is an example of a manuscript and related information stored in the manuscript storage unit according to the embodiment of the present invention. This is an example of a rule stored in the confirmation rule storage unit according to the embodiment of the present invention. This is an example of setting items stored in the setting information storage unit according to the embodiment of the present invention. It is a figure which shows an example of the flowchart of the process which displays the voice recognition result which concerns on embodiment of this invention for a proofreader. It is a figure which shows an example of the flowchart of the process of giving confirmation information to the voice recognition result which concerns on embodiment of this invention. This is an example of display information to which the voice recognition result and confirmation information according to the embodiment of the present invention according to the embodiment of the present invention are added. (Proofreader screen) It is an example of a display screen for a proofreader according to the embodiment of the present invention according to the embodiment of the present invention (proofreader screen). This is an example of a screen for displaying the voice recognition result according to the embodiment of the present invention according to the embodiment of the present invention as subtitles (screen of a television or the like).

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

FIG. 1 is a diagram showing an example of a system configuration according to an embodiment of the present invention. The system according to the embodiment of the present invention includes a voice recognition server 101 and an information processing terminal 102 (referred to as a speaker 102a, a proofreader 102b, and a display 102c).

The user inputs voice through the microphone 104 connected to the information processing terminal 102a. The information processing terminal 102a transmits the voice to the voice recognition server 101, converts it into a character string, sends it to the information processing terminal 102b (for the proofreader), displays it on the information processing terminal 102b (for the proofreader), and presents it to the proofreader. do.

The information processing terminals 102a to 102c may have both input and output of voice input and character string output. In the information processing terminal 102 output here, the display 102c and the proofreader 102b, which will be described later, may be combined, or each may be a dedicated information processing terminal. Further, the output is performed on the display device connected to the information processing terminal 102, but the configuration using a projector or the like is also the system configuration according to the embodiment of the present invention. When using a projector, the information processing terminal 102 is only one for the speaker, and all may read the character string of the voice recognition result displayed on the screen from the projector connected to the information processing terminal 102a. In that case, the speaker himself or another user may calibrate the misrecognition directly on the information processing terminal 102a for the speaker as a proofreader.

Further, the voice recognition server 101 may exist on the cloud, and in that case, the user of this system may use the function on the voice recognition server 101 described later by the cloud service. good. That is, the voice recognition unit 322, which will be described later, may be a function on another server called from the voice recognition server 101 or a service on the cloud. That is, even if these services are used as other servers or cloud services, the system configuration according to the embodiment of the present invention is used.

The information processing terminals 102a to 102 described in the configuration example have both input and output, but may be separated into input-only and output-only.

The voice recognition server 101 and the information processing terminals 102a to 102c may be in the same housing. That is, software capable of voice recognition may be installed in one of the information processing terminals 102a to 102 in FIG. 1 and may also serve as the voice recognition server 101.

FIG. 2 is a block diagram showing an example of a hardware configuration applicable to the voice recognition server 101 and the information processing terminal 102 according to the embodiment of the present invention.

As shown in FIG. 2, the information processing server 100, the recognition server 101, and the information processing terminal 102 are the CPU (Central Processing Unit) 201, the RAM (Random Access Memory) 203, and the ROM (Read Only Memory) via the system bus 204. A configuration is adopted in which 202, an input controller 205, a video controller 206, a memory controller 207, a communication I / F controller 208, and the like are connected.

The CPU 201 comprehensively controls each device and controller connected to the system bus 204.

Further, the ROM 202 or the external memory 211 will be described later, which is necessary for realizing the functions executed by the BIOS (Basic Input / Output System) and the OS (Operating System), which are the control programs of the CPU 201, and the functions executed by each server or each PC. Various programs etc. are stored. In addition, information necessary for carrying out the present invention is stored. The external memory may be a database.

The RAM 203 functions as a main memory, a work area, and the like of the CPU 201. The CPU 201 realizes various operations by loading a program or the like necessary for executing the process from the ROM 202 or the external memory 211 into the RAM 203 and executing the loaded program.

Further, the input controller 205 controls input from a pointing device such as a keyboard (KB) 209 or a mouse (not shown).

The video controller 206 controls the display on a display such as the display 210. The display may be a display such as a liquid crystal display. These are used by the administrator as needed.

The memory controller 207 is an external storage device (hard disk (HD)) for storing boot programs, various applications, font data, user files, edit files, various data, etc., a flexible disk (FD), or a PCMCIA (Personal Computer). Controls access to external memory 211 such as CompactFlash® memory connected via an adapter to the Memory Card International Association card slot.

The communication I / F controller 208 connects and communicates with an external device via the network, and executes communication control processing on the network. For example, communication using TCP / IP (Transmission Control Protocol / Internet Protocol) is possible.

The CPU 201 can be displayed on the display 210, for example, by executing an outline font expansion (rasterization) process in the display information area in the RAM 203. Further, the CPU 201 enables a user instruction by a mouse cursor (not shown) or the like on the display 210.

Various programs described later for realizing the present invention are recorded in the external memory 211, and are executed by the CPU 201 by being loaded into the RAM 203 as needed.

FIG. 3 is a diagram showing an example of a functional configuration according to an embodiment of the present invention. As described with reference to FIG. 1, the information processing terminal 102 may have the functions of the speaker 102a, the proofreader 102b, and the display 102c in different terminals or in a common terminal. Here, each will be described without distinction.

The voice acquisition unit 311 inputs utterances by the speaker's voice from a microphone or the like built in or connected to the information processing terminal 102 as voice data, and transmits the voice data transmission unit 312 to the voice recognition server 101.

The voice recognition server 101 passes the voice data received by the voice data receiving unit 321 to the voice recognition unit 322, converts the voice data into a character string, and presents the character string to the calibrator by the confirmation information addition unit 323. Is given. At that time, the manuscript stored in the manuscript storage unit 331 (detailed in FIG. 4) and the rules and patterns stored in the confirmation rule storage unit 332 (detailed in FIG. 5) are referred to. Further, the result of adding the confirmation information is sent to the information processing terminal 102 by the confirmation information transmission unit 324.

The confirmation information receiving unit 313 of the information processing terminal 102 receives the result of voice recognition to which the above-mentioned confirmation information is added, displays it on the display device connected to the information processing terminal 102 by the calibration display unit 314, and displays it on the calibration unit. The 315 accepts an operation by the calibrator and sends the calibration result to the voice recognition server 101 by the calibration result transmission unit 316.

The calibration result reception unit 325 of the voice recognition server 101 receives the calibration result, generates subtitles based on the voice recognition result, confirmation information, and calibration result, and the subtitle distribution unit 326 transmits the subtitles for display.

The information processing terminal 102 receives the subtitles transmitted by the subtitle distribution unit 326 by the subtitle receiving unit 216 and displays them on the display device. Actually, it is not necessary to display it on the information processing terminal 102. For example, even if it is further transmitted to a device for superimposing news or weather forecast images, or in the case of television broadcasting, it is transmitted to a device for distributing images to homes, etc. good. That is, it goes without saying that the final display destination of the subtitles received by the subtitle receiving unit 216 may be further transmitted (delivered) from the corresponding information processing terminal 102.

FIG. 4 is an example of a manuscript and related information stored in the manuscript storage unit according to the embodiment of the present invention. In this figure, the manuscript example 400 is a manuscript to be read aloud by the speaker (announcer or the like). The sentences included in this manuscript are separated by some standard in advance and stored in the manuscript storage unit 331.

The criteria for dividing the text of the manuscript are not particularly limited in the present invention, and no rules are shown. However, well-known techniques include the number of morphemes when morphological analysis is performed, the length of a character string, and the like. Generally, it is sufficient if the length is such that a significant similarity can be calculated as compared with the speech recognition result.

The manuscript storage unit 331 stores the unit for dividing the sentence as one line (content 405). The manuscript number 401 indicates which manuscript each line belongs to, and the plurality of lines in which the manuscript shown in the manuscript 2 (manuscript example 400) is stored are collectively referred to as "2" in the manuscript number 401. It is described as. Information on the manuscript corresponding to the manuscript numbers 1 to 5 is registered in the manuscript storage unit 331 of FIG.

Further, the rule number 402 is also summarized in the same manner, but the confirmation rule corresponding to the manuscript number 2 indicates that the rule number 501 “1” in FIG. 5 described later corresponds to the rule number 501, and the proofreading is performed according to the rule of this number. Confirmation information for the person will be given. Further, the time 403 is a time when the manuscript is supposed to be read aloud. Actually, there may be a gap, but if there are multiple similar news items, it is used to determine which news manuscript you are reading.

The information type 404 indicates whether the character string described in the content 405 is "inside the manuscript" or "outside the manuscript". "Inside the manuscript" indicates that it is obtained by separating from the manuscript example 400. On the other hand, "outside the manuscript" is not included in the manuscript read out by the announcer (manuscript example 400), but the summary of the news is given to viewers other than deaf people who read the voice recognition result as subtitles 1-2. It is shown by a line, and is displayed so as to be superimposed on the image on the screen, for example, as shown in the non-manuscript display character string 1102 in FIG. In the present invention, it is distinguished from the subtitles transcribed from the voice recognition result.

The state 406 is a description indicating whether or not each line has already been read aloud by an announcer or the like. Three types of "completed", "in use", and "unused" are listed, but even if it is "completed", it does not make a strict judgment that the same utterance will never be made after that. .. The announcer may read the part of the manuscript once again at his own discretion (for example, while the reporter in the field is not connected). However, if it is "completed", the probability that it will be read aloud is still low. It is information that can be used as a reference for the judgment of.

FIG. 5 is an example of a rule stored in the confirmation rule storage unit 332 according to the embodiment of the present invention. Even if there is a difference between the manuscript and the voice recognition result, not all the information is important, so from the viewpoint of emphasizing real-time performance, identify the important part from the character string of the voice recognition result. , Information that prompts confirmation (confirmation information) is given and can be presented to the proofreader. In the present embodiment, the rule is applied to the character string of the voice recognition result provided to the proofreader and the confirmation information is given, but in the actual processing, it may be given to the manuscript side. When it is given to the manuscript side, some information can be given in advance when the manuscript is registered in the manuscript storage unit 331, not at the same time as the voice recognition result. For example, information on part of speech and attached words, which will be described later, may be added to the manuscript side in advance, and confirmation information may be transcribed on the character string side of the voice recognition result when compared with the voice recognition result. Any method may be used as long as it can be provided to the information processing terminal 102b (for the proofreader).

I will explain by raising some rules, but it goes without saying that this embodiment is just a technical example, and it is necessary to create appropriate rules according to the actual news content and the situation at the site.

The rule number 501 is an identification number for associating with the rule number 402 in FIG. The identification display condition 502 specifies whether to prompt the proofreader to confirm only when the corresponding portion of the document and the voice recognition do not match (only the mismatch), or whether to prompt the confirmation even if they match (including the match). For example, if a person's name is mistaken in the news, it may be related to human rights issues, so it is conceivable to check with the proofreader even if the manuscript and the voice recognition result match. Even though the manuscript and the voice recognition result match, the registered manuscript is just multiple news items, and it is possible that the voice recognition result matches the other parts of the manuscript. Is.

The application rule 503 is associated with the rule name 504 of the rule details 500 described later. For example, in the description of "numerical expression (date, time, amount of money)", the rule name 504 of the rule detail 500 includes all of "numerical expression (date)", "numerical expression (time)", and "numerical expression (amount)". Means apply to strings.

The pattern 505 extracts a characteristic expression from a manuscript or a character string of a voice recognition result by using part-of-speech information of a morpheme, a character pattern, and the like. Some speech recognition services give part of speech to each morpheme. Alternatively, the resulting character string may be given a part of speech by morphological analysis in the processing of this embodiment. The character pattern is, for example, a character type (Kanji, Hiragana, Katakana, numerical value, etc.), a prefix of a person's name, a suffix, or the like. In addition, a prefecture name or the like may be registered in a dictionary (not shown) and a character string may be matched. A method of specifying these part of speech and character patterns by using a regular expression or the like is a well-known technique (for example, Japanese Patent Application Laid-Open No. 2001-125911).

In addition, ancillary words may be important as items to be confirmed by the proofreader. If the auxiliary verb of cancellation is wrong, it will tell the opposite fact completely, and if the tense is wrong, it will be misunderstood. Even if it is not so much, expressions such as "..." may give a casual and inappropriate impression in criminal cases. On the other hand, in the case of interviews and weather forecasts, it is not so inappropriate to include casual expressions, and in some cases it is not necessary to check them in order to reduce the burden on the proofreader. For attached words, the degree of misunderstanding is determined from the order, and the "confirmation importance" (in the above-mentioned patent document, the "risk" of expressions such as claims is considered as the importance) is calculated. It is also a well-known technique (for example, Japanese Patent Application Laid-Open No. 2004-133714). The adjunct may be one or more sequences. In the case of a plurality of sequences, the sequence of part of speech and characters may be represented as a pattern by a regular expression or the like, as in the above-mentioned pattern. The permissible range of this risk may be described in the identification display condition 502, and if the risk exceeds the risk, it may be presented to the proofreader. Further, it may be specified as the adjoint word determination threshold value in FIG.

FIG. 6 is an example of setting items stored in the setting information storage unit according to the embodiment of the present invention. I will explain each setting.

If there is no input for a certain period of time in voice recognition (if no human voice is input), it is recognized as a sentence. The value that specifies the break time for that purpose is the utterance break time, which is 0.5 seconds in the example. For example, one sentence as a voice recognition result may be specified by this delimiter, and the character string may be compared with the "content" of FIG. 4 to determine whether or not they are similar. However, in this case, if the recognition regarding the delimiter between the manuscript creator and the speaker is different, it may not match the specific content of the manuscript storage unit 331 due to the positional deviation of the character string. Therefore, the number of morphemes specified by the similar comparison morpheme number or the length of the character string may be set. In Patent Document 1, four morphemes are always used as one line, and the same two morphemes in the latter half of one line and the next two morphemes are specified. That is, although it is shifted little by little, the same method may be used in the embodiment of the present invention. As the embodiment of the present invention, any method may be used as long as it is a storage method for determining whether or not they are similar.

Next, the manuscript confirmation range will be described. In the example, it is 10 lines. If the line currently in use in the manuscript storage unit 331 of FIG. 4 (the second line of the manuscript number 2) and the voice recognition result of the speaker is similar to the line separated by 10 lines or more, the manuscript It is judged that the position of is inappropriate.

If you proceed to the next line, you should get a recognition result similar to the sentence "from the bank account of a woman living in the city", but it is similar to "according to police investigation". In this case, the following are possible possibilities. If, for some reason, the manuscript 2 currently being read aloud is repeated from the beginning, or if the manuscript 2 is skipped and the manuscript 3 is moved to the manuscript 3, and the utterance corresponding to the third line is made, urgent news is entered, and the manuscript 2, the manuscript 2, For example, when the announcer is speaking news that is neither of 3 (not in the manuscript).

In this case, in order to confirm which is correct, information for confirming the manuscript number may be added to the voice recognition result and sent to the information processing apparatus 102b (for the proofreader). FIG. 10 shows the situation. The operation panel 1002 may be displayed on the proofreader operation screen 1000 together with the proofreading target character string 1001 so that the proofreader can confirm whether to switch to the manuscript 3. In the case of urgent news (without manuscript), the instruction "The proofreader should check everything" may be presented. Further, when switching of the original is selected, in the subsequent processing, the subsequent processing may be continued centering on the original after switching and its line.

In any case, it does not mean that the recognition result cannot be used if a certain standard is not met, but it allows the proofreader to judge how to use the recognition result.

Further, in the manuscript of FIG. 4, only numbers are described in the manuscript number 401, but a title whose contents can be understood may be given as shown in FIG.

Next, the flow of processing according to the embodiment of the present invention will be described with reference to the flowcharts of FIGS. 7 and 8.

In step S701, the voice input by the speaker is received from the microphone 104 connected to the information processing terminal 102a (for the speaker), and the voice data is transmitted to the voice recognition server 101 in step S702.

The voice recognition server 101 receives the voice data transmitted from the information processing terminal 102a in step S703, performs voice recognition processing in step S704, and converts it into a character string. At this time, a portion where the silent state of the continuously input voice data is longer than the "utterance break time" in FIG. 6 may be detected, the voice data may be divided by the silent state portion, and then passed to the voice recognition process. .. Further, some voice recognition engines output the result of automatic division in a silent state, and based on the result, the character string may be divided so as to correspond to one line of the document.

Step S705 will be described with reference to the flowchart of FIG. In step S801, it is determined from which step this flowchart is called. In the current description, since it is called continuously from step S704, the process proceeds to step S802.

In step S802, based on the character string as a result of voice recognition, which line content (character string) of the document storage unit 331 is similar is listed. Even if the manuscript 2 is focused at the present time, it may be listed from other parts such as the manuscript 3. Alternatively, a morphological analysis may be performed instead of a character string to list similar lines as a morpheme string. As will be described later, after completing the flowchart of FIG. 8, the flowchart of FIG. 7 may be returned and the proofreader may return to FIG. 8 again at the discretion of the proofreader. In that case, the call is made as a process following step S710, and the branch of step S801 proceeds to step S803. Therefore, step S802 is executed only once at the beginning. That is, the process of listing similar lines from the manuscript is only the first.

In step S803, among the similar lines of the manuscripts listed in step S802, the most similar lines are selected and set as the lines of interest. As described above, the flowchart of FIG. 8 may be called repeatedly, but in the second execution, the row selected in the first time is excluded from the candidates, so the most similar one should always be selected. Based on the one listed first, the first, second, and third place and the next candidate line are sequentially focused on according to the number of times the flowchart of FIG. 8 is called.

In step S804, the difference between the character string of the voice recognition result and the character string of the similar line listed from the manuscript is extracted, and the information indicating the difference is stored.

In step S805, the portion to be confirmed by the proofreader is extracted and the confirmation information is given according to the rule corresponding to the manuscript stored in the confirmation rule storage unit 332. Further, in step S806, the confirmation importance of the adjunct word (adjunct word string) to which the confirmation information is added is calculated, it is determined that the adjoint word determination threshold is exceeded, and the confirmation information is added if necessary.

In step S807, it is determined whether or not the line currently being focused on is at a position far from the line of the document completed immediately before. For example, even if the first line of the manuscript 3 is focused on as a similar line when the manuscript 2 is at the end, the progress is natural. This can be determined based on the document confirmation range (10 lines) in FIG. However, if the back of the manuscript 3 is focused on as a similar line when the manuscript 2 is at the beginning, there is a possibility of an error. Further, it is natural that the state 406 of the document storage unit 331 progresses to the "not yet" line, but it is not natural that the state 406 progresses to the "finished" line. Of course, the proofreader may actually read the manuscript again, and it is necessary for the proofreader to judge whether it is correct or not. .. Determine whether to assign the identifier. If it is determined that the grant is necessary, the process proceeds to step S808. If it is determined that it is not necessary to give it, the flowchart of FIG. 8 is terminated.

In step S808, the voice recognition result at the present time gives information that the range is very far from the line of the document completed immediately before and is out of the range. This completes the flowchart of FIG. 8 and returns to the state in which step S705 of the flowchart of FIG. 7 is completed.

Returning to the description of the flowchart of FIG. 7. In step S706, the voice recognition result and the given confirmation information are transmitted to the information processing terminal 102b (for the proofreader).

In step S707, the information processing terminal 102b (for the proofreader) receives the voice recognition result and the given confirmation information, and displays them on the display device connected so that the proofreader can work in step S708. The displayed example will be described later with reference to FIG.

In step S709, an operation such as a correction or a judgment performed by the proofreader on the voice recognition result based on the display information of FIG. 9 is accepted. This determination also includes a determination as to whether or not to allow switching of the manuscript already described with reference to FIG.

Step S710 is a determination for branching the subsequent processing flow when the determination of whether or not to switch the original is included as a result of the operation of step S709. If the proofreader approves the switching of the manuscript, the candidate line in the manuscript of interest in step S803 may be used, and the process proceeds to the subsequent step S711 (in the case of YES). If not, it is necessary to select and cure the candidate line, so the process is returned to step S705 of the voice recognition server 101 (in the case of NO). The explanation of step S705 is the same as that described above as the flowchart of FIG. 8, but since step S802 has already listed similar lines from the manuscript at the time of the first execution, it is skipped (it becomes "continuation of S710" at the branch of step S801). ), We will proceed with the processing focusing on the next rank from the candidate rows of similar rows.

In step S711, the calibration result (final display character string) performed by the calibrator in step S709 is transmitted to the information processing terminal 102c (for display).

In step S712, the information processing terminal 102c (for display) receives the calibration result (final display character string) and displays it on the display device connected to the information processing terminal 102c (for display) in step S713.

This completes the description of an example of the processing flow according to the embodiment of the present invention using the flowcharts of FIGS. 7 and 8.

FIG. 9 is an example of display information to which the voice recognition result and confirmation information according to the embodiment of the present invention according to the embodiment of the present invention are added (calibrator screen). For comparison, a character string of the manuscript 2 (manuscript example 400) and an example of the voice recognition result (voice recognition result 901) are shown. In the voice recognition result 901, the portion surrounded by the rectangle is the difference from the original example 400.

As for the display for the proofreader, a character string corresponding to each line of the manuscript 2 in FIG. 4 is displayed from 902a to e as in the display for the proofreader 902. Since it is received and displayed when the recognition of one line is completed, all of these lines are not displayed at the same time but are displayed one by one (see the proofreader operation screen 1000 in FIG. 10). 902a to e will be explained one by one.

In 902a, since the voice recognition result and the document match, only the character string is displayed.

In 902b, the difference between the manuscript and the voice recognition result is in two places, 921 and 922, and the proper nomenclature (personal name) and numerical expression (date) of rule number 1 in FIG. 5 are applied, so that the proofreader can identify them. It is highlighted as much as possible (solid rectangle), and the correct answers ("Inagawa" and "July", respectively) included in the manuscript side are displayed in balloons. Here, after the proofreader confirms, the character string in the voice recognition result can be replaced with the character string in the balloon by an operation such as clicking the balloon portion with the mouse. Alternatively, the character string of the voice recognition result may be edited directly. Also, if the character string of the voice recognition result is correct for some reason, there is no need to replace it.

The proper noun (personal name) of 921 is given confirmation information in order to "make the proofreader confirm even if they match" in the rule of FIG. If they match, the callout does not have to be displayed, but the solid rectangle is displayed.

In addition, "September" in 923 matches the manuscript, and the numerical expression (date) is to be presented to the proofreader only when the numerical expression (date) does not match in the rule of FIG. Only the matching parts may be displayed in an identifiable manner (dotted rectangle).

In 902c 924, since the "account" of the manuscript is a "course" in the voice recognition result, the confirmation information is displayed by a solid rectangle. This is because there is a rule in FIG. 5 to give confirmation information if there is a discrepancy even if it is a common noun. If the importance is low and the work efficiency of the proofreader is important, it is not necessary to delete it from the rule of FIG. 5 and display it. Even in that case, since there is a difference from the original, the proofreader may be urged to confirm the proofreader at a lower level by adding a dotted rectangle.

Since 902d has no difference from the original, the confirmation information is not displayed. However, confirmation information at a lower level may be displayed in "700,000 yen" (numerical information (amount)).

There are two differences in 902e. 926 "no" is not displayed because it is not included in the rule of FIG. The "done" in 927 is "su" in the manuscript, and the difference in tense between the present tense and the past tense causes a great misunderstanding for the reader of the subtitles, so the confirmation information is displayed as a solid rectangle. ..

As described above, the calibrator can calibrate in a short time while confirming this information. This completes the description of FIG.

FIG. 11 is an example of a screen for displaying the voice recognition result according to the embodiment of the present invention as subtitles (screen of a television or the like). For example, a scene in which an announcer is reading a manuscript is displayed on a TV news program.

The voice recognition result character string 1101 receives the result of completion of calibration from the information processing terminal 102b and displays it as it is. ">>" is added at the beginning of the line to clearly indicate that it is a subtitle of the voice recognition result.

Further, on a screen such as news, the voice of the announcer of the present invention is often displayed as a large telop, such as a summary of news, in addition to subtitles using the voice recognition result. In order to distinguish this from the subtitles of the voice recognition result, it is referred to as the character string 1102 displayed outside the manuscript.

When both the voice recognition result character string 1101 and the non-manuscript display character string 1102 are displayed, there are problems such as the characters overlapping and becoming very difficult to read, and the large number of characters cannot be read in a short time. Occur.

Therefore, a method of not displaying the voice recognition result will be described. In the last line of the manuscript number 2 of the manuscript storage unit 331 of FIG. 4, "Suspect Inagawa, suspected embezzlement in business" is registered as "outside the manuscript". The term "outside the manuscript" means that the character string 1102 outside the manuscript of FIG. 11 is determined to be displayed regardless of the announcer's utterance.

If this non-manuscript display character string 1102 is displayed, sufficient information is transmitted to the viewer even if the voice recognition result corresponding to the line "suspicion of embezzlement in business" is not displayed in the manuscript. Conceivable. Therefore, an additional explanation will be given to the flowcharts of FIGS. 7 and 8.

When a line similar to the voice recognition result is listed in step S802 of FIG. 8, if the line "outside the manuscript" is included in the higher rank or with a very high degree of similarity, the display outside the manuscript is displayed on the television. It is conceivable that the process for the voice recognition result is interrupted, that is, the process returns to step S701 and the process for the next voice input is started. As a result, the corresponding speech recognition result is not sent to the subsequent processing and does not exist.

Alternatively, after the confirmation information that "the display outside the manuscript exists" is given in the process of step S802, the same process as the conventional description is continued, and the proofreader is informed that "the display outside the manuscript exists". It is also possible to display and not display at the discretion of the proofreader.

In any case, it is possible to automatically decide not to display the subtitles or by the proofreader with a simple operation, and it is possible to obtain the effect of enabling more real-time subtitle display. In addition, it is possible to obtain the effect of avoiding having to read a large amount of characters on the screen from the viewer's point of view. This completes the description of FIG.

Although some embodiments have been described above, the present invention can be, for example, an embodiment as a system, an apparatus, a method, a computer program, a recording medium, or the like, and specifically, a plurality of devices. It may be applied to a system composed of, or may be applied to a device consisting of one device.

Further, the computer program in the present invention is a computer program in which a computer can execute the processing methods shown in FIGS. 7 and 8, and the storage medium of the present invention can execute the processing methods in FIGS. 7 and 8. Computer programs are stored. The computer program in the present invention may be a computer program for each processing method of the devices of FIGS. 7 and 8.

As described above, a computer in which a recording medium on which a computer program that realizes the functions of the above-described embodiment is recorded is supplied to the system or device, and the computer (or CPU or MPU) of the system or device is stored in the recording medium. Needless to say, the object of the present invention is achieved by reading and executing the program.

In this case, the computer program itself read from the recording medium realizes the novel function of the present invention, and the recording medium storing the computer program constitutes the present invention.

Recording media for supplying computer programs include, for example, flexible disks, hard disks, optical disks, magneto-optical disks, CD-ROMs, CD-Rs, DVD-ROMs, magnetic tapes, non-volatile memory cards, ROMs, and EEPROMs. Silicon disks, solid state drives, etc. can be used.

Further, by executing the computer program read by the computer, not only the function of the above-described embodiment is realized, but also the OS (operating system) or the like running on the computer is realized based on the instruction of the computer program. Needless to say, there are cases where a part or all of the actual processing is performed and the processing realizes the functions of the above-described embodiment.

Further, the computer program read from the recording medium is written to the memory provided in the function expansion board inserted in the computer or the function expansion unit connected to the computer, and then its function is based on the instruction of the computer program code. Needless to say, there are cases where the CPU provided in the expansion board or the function expansion unit performs a part or all of the actual processing, and the processing realizes the functions of the above-described embodiment.

Further, the present invention may be applied to a system composed of a plurality of devices or a device composed of one device. It goes without saying that the present invention can also be applied when it is achieved by supplying a computer program to a system or an apparatus. In this case, by reading the recording medium containing the computer program for achieving the present invention into the system or device, the system or device can enjoy the effect of the present invention.

Further, by downloading and reading a computer program for achieving the present invention from a server, database, or the like on a network by a communication program, the system or device can enjoy the effect of the present invention. It should be noted that the present invention also includes all the configurations in which each of the above-described embodiments and modifications thereof are combined.

101 Voice recognition server 102 Information processing terminal 311 Voice acquisition unit 312 Voice data transmission unit 313 Confirmation information reception unit 314 Calibration display unit 315 Calibration unit 316 Calibration result transmission unit 317 Subtitle reception unit 321 Voice data reception unit 322 Voice recognition unit 323 Confirmation Information addition unit 324 Confirmation information transmission unit 325 Calibration result reception unit 326 Subtitle distribution unit 331 Manuscript storage unit 332 Confirmation rule storage unit

Claims (8)

An information processing device that acquires recognition data that is a voice recognition result of voice data and text data corresponding to the voice data.
A specific means for identifying a point of interest in the recognition data or the text data, and
An information processing apparatus including a display control means for identifying and displaying the specified portion when displaying the result of comparing the recognition data and the text data. The information processing apparatus according to claim 1, wherein the specifying means identifies the location according to an attribute of a character string included in the data. The information processing according to claim 1 or 2, wherein the display control means controls the display method of the location based on the result of comparing the recognition data and the text data at the specified location. Device. The information according to claim 1 to 3, wherein the display control means notifies the data of the text data at the specified location when there is a difference between the recognition data and the text data. Processing equipment. The information processing apparatus according to any one of claims 1 to 4, further comprising a receiving means for receiving correction of the recognition data at the specified location. The information processing apparatus according to any one of claims 1 to 5, wherein the text data is text data of a manuscript of the voice data. It is a control method of an information processing device that acquires recognition data that is a voice recognition result of voice data and text data corresponding to the voice data.
A specific step in which the specific means identifies a point of interest in the recognition data or the text data, and
A control method for an information processing apparatus, wherein the display control means includes a display control step for identifying and displaying the specified portion when displaying a result of comparing the recognition data with the text data. A program that can be executed in an information processing device that acquires recognition data that is a voice recognition result of voice data and text data corresponding to the voice data.
The information processing device
A specific means for identifying a point of interest in the recognition data or the text data, and
A program for functioning as a display control means for identifying and displaying the specified portion when displaying the result of comparing the recognition data and the text data.

Images