JP5184071B2 - Transcription text creation support device, transcription text creation support program, and transcription text creation support method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an apparatus and a program for supporting audio documentation, significantly reducing the risk of information leak. <P>SOLUTION: A voice division device 2 divides voice data in a voice data storage part 1 to relatively short voice sections. An identifier assignment device 3 performs voice recognition to each divided voice file and collates each file using a keyword list for confidential information. A documentation control device 4 requests documentation to staff with high reliability when the recognition result of the voice file contains a keyword, and requests documentation to general staff when the recognition result contains no keyword for confidential information. A text extraction device 5 extracts a documentation text from a documentation result and stores it in a text data storage part 6. <P>COPYRIGHT: (C)2009,JPO&amp;INPIT

Description

  The present invention relates to a transcription text creation support apparatus and a transcription text creation support program.

  Conventionally, so-called transcription work for converting voice data into text data is required in various fields. For example, the minutes are created from the recording of the meeting, or the recording of the dialogue between the user and the operator in the call center is made into text and used for information analysis. The conventional transcription work mostly relies on human manual work, but in recent years, with the advancement of speech recognition technology by computers, some automatic text creation by speech recognition has been introduced.

However, considering the current level of speech recognition technology, there are still many misrecognitions, so it is almost impossible to transcribe with high accuracy by speech recognition. Therefore, when using speech recognition in the transcription of speech data, first, the speech data is converted into text by speech recognition by a computer, and then the result of speech recognition is checked and corrected manually. Is common.
For example, voice information is recognized and converted into text, the text information is divided into small units, the text information is displayed on the screen in divided units, and the text converted into text in divided units is checked. A technique for updating text has been proposed (see, for example, Patent Document 1).

Regardless of whether or not speech recognition is introduced, transcription of speech data requires a large amount of manual work, and thus requires a lot of work staff. However, since the amount of work that can be entrusted varies greatly depending on the time for a contractor who undertakes transcription work, all work cannot be relied on by dedicated staff such as full-time employees. Usually, a large number of staff members who can be operated are registered in advance, and only when necessary, the registered staff members are mobilized according to the amount of work to perform work.
JP 2005-228178 A

  By the way, in the above-described prior art, there may be a situation in which a large amount of non-exclusive staff has to be relied upon in cost for transcription of a huge amount of audio data. Here, when confidential information and privacy information are mixed in the audio data, there is a variation in staff history, ability, and confidentiality, so consider the risk of information leakage via the staff. It is desirable.

  Conventional transcription technology has dealt with the risk of information leakage by taking measures such as letting the staff write a pledge or seeking damages in the event of information leakage. However, such measures may not be able to cope with information leakage by moral or malicious staff.

  In addition, in the present situation where diverse and large numbers of staff are involved in the transcription work, it is necessary to take different measures for each staff in order to protect the confidentiality of information. Furthermore, how to deal with audio data varies depending on the presence or absence of confidential information. In other words, it is desirable that the system and mechanism be capable of flexibly responding to the situation of staff and audio data.

  However, the conventional technique has a problem in that it cannot perform fine work distribution according to the reliability of staff confidentiality according to confidential information included in audio data.

  The present invention has been made in consideration of such circumstances, and an object thereof is to provide a transcription text creation support apparatus and a transcription text creation support program capable of greatly reducing the risk of information leakage. There is.

In order to solve the above-described problems, the present invention is a transcription text creation support apparatus for transcription of text data from speech data, and each transcription worker has a reliability of each transcription worker. Reliability storage means for storing the corresponding reliability value, divided sound data generation means for generating divided sound data by dividing the sound data into predetermined sections, and divided sound data divided by the divided sound data generation means A conversion means for performing speech recognition every time and converting it into a plurality of primary text data, a keyword storage means for storing a keyword relating to confidential information, and storing each primary text data converted by the conversion means in the keyword storage means Based on the determination result by the determination means and the determination result by the determination means A selection unit that refers to the reliability value of the reliability storage unit and selects a transcription worker to which a transcription operation should be requested for each divided voice data divided by the divided voice data generation unit. Is a transcription text creation support device.

  According to the present invention, in the above invention, the transmission means for transmitting the corresponding divided voice data to the terminal of the transcription operator selected by the selection means, and the text data of the result of the transcription work by the transcription operator are received. And receiving means for integrating the text data received by the receiving means and storing the data as transcription text data for the original voice data.

  According to the present invention, in the above invention, when the selection unit refers to a reliability value of the reliability storage unit and the determination unit determines that the primary text includes a keyword related to confidential information. Select a transcription worker with a high reliability value, and if it is determined that the primary text does not contain a keyword related to confidential information, select a transcription worker with a low reliability value. Features.

In order to solve the above-described problem, the present invention is a transcription text creation support program for supporting a transcription work for transcribing text data from voice data. Generating divided audio data by dividing the data into a plurality of steps, a step of performing speech recognition for each of the divided divided audio data and converting it into a plurality of primary text data, and a keyword related to confidential information for each converted primary text data And when it is determined that a keyword related to confidential information is included in the primary text, a transcription operator having a high reliability value is selected, and the primary text is selected. If it is determined that keywords related to confidential information are not included, the transcription worker with a low reliability value A text creation support program transcription, characterized in that and a step of selecting.
In order to solve the above-described problem, the present invention provides a transcription text creation support method for supporting a transcription work for transcription of text data from voice data , which is executed by a computer, and which is divided voice data. A generating unit that divides the audio data into predetermined sections to generate divided audio data; and a conversion unit that performs speech recognition for each of the divided divided audio data and converts the audio data into a plurality of primary text data. determination means, for each transformed primary text data, and determining whether the keyword is included about confidential information, the selection means is determined to contain a keyword relating to confidential information to the primary text If a transcriber with a high reliability value is selected, a keyword related to confidential information is selected in the primary text. If it is determined that no rare, is transcribed text creation support method characterized by comprising the steps of confidence value is selected lower transcription worker.

  According to the present invention, voice data is divided into predetermined sections to generate divided voice data, voice recognition is performed for each divided voice data, converted into primary text data, and each converted primary text data is converted. Whether or not a keyword related to confidential information is included, and based on the determination result, for each divided voice data divided by the divided voice data generation means, a transcription worker to be requested for the transcription work is selected. To do. Therefore, it is possible to obtain an advantage that the risk of information leakage can be greatly reduced by dividing voice data and requesting a staff member with high reliability to transcribe a portion including confidential information.

  Further, according to the present invention, the divided voice data is transmitted to the selected transcription worker's terminal, the text data of the transcription work result by the transcription worker is received, and the received text data is integrated. , And save it as a transcript text data for the original audio data. Therefore, by dividing the voice data and requesting it from different staff, the risk of information leakage can be further reduced, and there is an advantage that the transcript text data for the original voice data can be finally obtained. .

  According to the present invention, the reliability value is high when the reliability value of the reliability storage unit is referred to and the determination unit determines that the primary text includes a keyword related to confidential information. A transcription worker is selected, and if it is determined that the primary text does not contain a keyword related to confidential information, a transcription worker having a low reliability value is selected. Therefore, it is possible to obtain an advantage that the risk of information leakage can be greatly reduced by dividing voice data and requesting a staff member with high reliability to transcribe a portion including confidential information.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing a configuration of a transcription text creation support apparatus according to an embodiment of the present invention. In the figure, a voice data storage unit 1 stores voice data to be transcribed. The voice dividing device 2 reads the voice data from the voice data storage unit 1, divides the voice data into relatively short voice sections (for example, sentence units) using the silent section during speech, and divides the voice data into voice sections. Save the audio data as individual files.

  The audio dividing device (divided audio data generating means) 2 creates a management table for managing audio files (hereinafter referred to as divided audio files) in each audio section, and generates file names of the divided audio files. The file is stored in the management table in association with the divided audio file. Details of the management table will be described later.

  Next, the identifier assigning device (conversion unit) 3 performs speech recognition on each divided speech file and stores the recognition result text in the management table. Further, the identifier assigning device 3 collates the recognition result using a preset keyword list (corresponding to confidential information), and the confidential information is included in the divided audio file including at least one keyword. An identifier indicating that is assigned.

  The transcription control device 4 allocates each divided audio file to the client terminals (such as PCs of work staffs who request the transcription work) 8-1 to 8-n via the network 7 and transmits them to the transcription work. And the work results are received and stored from the client terminals 8-1 to 8-n.

  Regarding the working staff, the reliability in confidentiality is evaluated in advance, and classified into highly reliable staff and general staff. In the allocation to the client terminals (work staff) 8-1 to 8-n, only the staff with high reliability is requested to work on the divided audio file to which the identifier of the confidential information is assigned. You can ask either highly reliable staff or general staff.

  Next, the text extraction device 5 extracts text from the work result of each divided audio file received from the client terminal, connects it, and generates text data corresponding to the original audio data before division. The text data storage unit (storing means) 6 includes a storage device such as a hard disk, and stores linked transcription text data corresponding to the original voice data before division. The network 7 includes the Internet, a LAN, and the like. The client terminals 8-1 to 8-n are work staff terminals (PCs, etc.) that perform the transcription work.

  Next, FIG. 2 is a block diagram showing a configuration of the audio dividing device 2 according to the present embodiment. In the figure, a management table generation unit 2-1 generates a management table for storing and managing information in operation in a unified manner. The silent section threshold value calculation unit 2-2 first calculates a threshold value in order to identify the silent section and the voice section using the power value (magnitude) of the voice data. Since there is usually a silent section from several tens of milliseconds to several hundred milliseconds at the very front end of the audio data, the threshold is calculated using the silent section at the front end.

  The voice dividing unit 2-3 sequentially calculates the power of the voice data in units of frames from the beginning (starting end), and if a section smaller than the threshold continues for a certain amount, for example, 300 milliseconds, this section is set as a silent section. In digital signal processing, a speech waveform is normally sequentially processed in units of blocks, and this block is called a frame. The length of this frame is often 20 to 40 milliseconds.

  The audio dividing unit 2-3 cuts out audio data from the beginning to the midpoint of the silent section, and saves it as a divided audio file in the divided audio file storage unit 2-5. At this time, the voice dividing unit 2-3 writes the file name of the divided voice data in the management table, and simultaneously assigns a unique index number that is a serial number, and sets the “confidentiality flag”, “staff number”, “ Three items of “work status flag” are initialized. When the first segmentation is completed, the voice dividing unit 2-3 repeats the same operation for the remaining voice data, and thereby divides all input voice data into divided voice data of a relatively short voice section. . The file name and index of the divided audio data are stored in the management table. The management table storage unit 2-4 stores a management table (to be described later) that holds various types of information related to the divided audio file. The divided audio file storage unit 2-5 stores the divided audio file to which a unique file name is assigned.

  FIG. 3 is a conceptual diagram showing the data configuration of the management table 2-4-1 according to the present embodiment. The management table 2-4-1 includes items of “document number”, “index”, “file name”, “recognition result”, “transcription”, “confidentiality flag”, “staff number”, and “work status flag”. Is stored. The “document number” is a unique number assigned to an audio file (before division) of a predetermined unit that performs a transcription work. “Index” is a unique number that is a serial number. “File name” is the name of the divided audio file. The “recognition result” is a result (text) of speech recognition performed on each divided speech file. “Transcription” is the text of the corresponding divided audio file created by the transcription work performed by the operation staff operating the client terminal.

  The “confidentiality flag” is a flag indicating whether or not confidential information is included in the corresponding divided audio file. The initial value (non-confidential) is “0”, and is set to “1” when it is determined that there is confidential information. rewrite. The “staff number” is a number for identifying a staff member who is responsible for the transcription work of the divided audio file. The initial value is “0”, and the staff number is written when the staff in charge of the divided audio file is determined. The “work status flag” is a flag indicating whether or not all work of the corresponding divided audio file has been completed. The initial value (not finished) is “0”, and is rewritten to “1” when all work is completed. .

  Next, FIG. 4 is a block diagram showing a configuration of the identifier assigning device 3 according to the present embodiment. In the figure, a voice recognition unit 3-1 performs voice recognition on a divided voice file in the divided voice file storage unit 2-4 using a voice recognition engine that supports large vocabulary free speech, and recognizes each divided voice file. The text as a result is stored in the “recognition result” of the corresponding divided audio file in the management table 2-4-1 of the management table storage unit 2-4.

  The keyword storage unit (keyword storage unit) 3-2 stores a keyword table (described later) in which keywords that are confidential information are stored. The keyword collation unit (determination unit) 3-3 assigns one keyword from the keyword storage unit 3-2 to the recognition result of the first record in the management table 2-4-1 of the management table storage unit 2-4. When the collation is established, the collation for the corresponding divided audio file is terminated, and the “confidentiality flag” in the management table 2-4-1 is rewritten from “0” to “1”. At this time, if collation is not established even after collation with all keywords, collation with respect to the divided audio file is terminated as it is. In addition, after the collation with respect to the first divided voice file is completed, the keyword collation unit 3-3 performs collation with respect to all the divided voice files remaining sequentially in the same procedure.

  Next, FIG. 5 is a conceptual diagram showing an example of the keyword table 3-2-1 according to the present embodiment. The keyword table 3-2-1 includes “Index” and “Keyword” for identifying each keyword. The keyword list is provided by the user who is the transcription requester, or is created according to the instruction of the user who is the requester. The “keyword” usually includes personal information, confidential business information, and the like. In the example shown in FIG. 5, an individual name, age, sex, address, and the like are registered as “keyword”.

  Next, FIG. 6 is a block diagram showing the configuration of the transcription control device 4 according to the present embodiment. In the figure, the staff information acquisition unit 4-1 acquires the work status (busy / standby) and the staff number of the work staff from the client terminals 8-1 to 8-n, and stores the work status as the work staff information storage unit 4- 2 is stored in the work staff information table (described later). Further, the staff information acquisition unit 4-1 updates this information at regular time intervals (for example, 30 minutes).

  Here, FIG. 7 is a conceptual diagram showing a data configuration example of the work staff information table 4-2-1 according to the present embodiment. In the work staff information table (reliability storage means) 4-2-1, “staff number” is a number for identifying the work staff, and a number other than “0” is used. “Name” is the name of the work staff. The “reliability” is classified into two types, “general” and “high”, and is determined in advance based on the work staff's career and results. The “reliability” is usually divided into two types, “general” and “high”, but may be divided into three or more types as necessary. The “work status” indicates the current status (busy / standby) of the work staff and is acquired from the client. “Contact” is the contact information of the work staff, and stores, for example, an e-mail address.

  Next, returning to the description of FIG. 6, the management table information acquisition unit 4-3 becomes “0” in the “staff number” in the management table 2-4-1 of the management table storage unit 2-4. The divided audio file having the smallest “Index” is selected as the transmission target from the divided audio files (not requested for work). Then, information transmitted to any one of the client terminals 8-1 to 8-n is extracted from the management table 2-4-1 and the divided audio file storage unit 2-5 of the management table storage unit 2-4. At a minimum, “Index”, “file name”, and the divided audio file corresponding to the “file name” are transmitted to any of the client terminals 8-1 to 8-n, but the “recognition result” is also transmitted. By doing so, it is possible to easily create a transcription text by confirming and correcting the recognition result. That is, since the work of the staff can be carried out more efficiently, the text of the recognition result stored in the “recognition result” is usually transmitted.

  The transmission unit (selection unit, transmission unit) 4-4 refers to the “confidentiality flag” of the selected divided audio file from the management table 2-4-1 of the management table storage unit 2-4, and performs the work of the transmission destination. The transmission information (“Index”, “file name”, divided audio file, “recognition result”) extracted by the management table information acquisition unit 4-3 is assigned to any of the client terminals 8-1 to 8-n. Send. Specifically, the “reliability” and “work status” in the work staff information table 4-2-1 are referred to. When the “confidentiality flag” is “0”, the “reliability” is “general”. ”, The staff whose“ working status ”is“ standby ”is preferentially selected, the“ staff number ”of the selected waiting staff is stored in the“ staff number ”of the management table 2-4-1, and the above transmission is performed. The information is transmitted to the “contact” mail address of the staff of the “staff number” by e-mail, for example.

  On the other hand, when the “confidentiality flag” is “1”, only the waiting staff whose “reliability” is “high” is selected, and the “staff number” of the waiting staff is selected as the management table 2-4-1. Save and send to “Staff Number”. If there is no standby staff with “high” “reliability”, transmission is paused, and transmission is resumed when a standby staff with “high” reliability appears. The management table information acquisition unit 4-3 and the transmission unit 4-4 are performed until the contents of the “staff number” in the management table 2-4-1 are all filled.

  The receiving unit (receiving means) 4-5 receives “Index” and the transcribed text from any of the client terminals 8-1 to 8-n that have finished the transcription work, and refers to the “Index”. The transcription text is stored in “transcription” of the management table 2-4-1. At this time, the “work status flag” in the management table 2-4-1 is rewritten from “0” to “1”.

  Next, FIGS. 8A to 8D are conceptual diagrams for explaining the state transition of the management table 2-4-1 according to the present embodiment. First, FIG. 8A shows the state of the management table 2-4-1 at the time when the audio division is finished. At this time, the divided audio file names are stored, and the “confidentiality flag”, “staff number”, and “work status flag” are reset to “0”. Next, FIG. 8B shows the state of the management table 2-4-1 at the time when an identifier is assigned according to the sensitivity of the audio file. In this case, the confidentiality flag of the divided audio file whose “Index” is “2” and whose “file name” is “sp2.wav” is “1”.

  FIG. 8C shows the state of the management table 2-4-1 after the divided audio file or the like has been transmitted to any of the client terminals 8-1 to 8-n. At this time, since the work staff who performs the transcription is determined for each divided audio file, the work staff who performs the transcription for each divided audio file in the management table 2-4-1 is indicated. "Staff number" is stored. FIG. 8D shows the state of the management table 2-4-1 after being received from any of the client terminals 8-1 to 8-n. In this case, for each divided audio file, a “work status flag (“ 1 ”is work completed)” indicating the status of the work and “transcription” which is the text of the corresponding divided audio file created by the transcription work. And are saved.

  Next, FIG. 9 is a block diagram showing a configuration of the text extraction device 5 according to the present embodiment. In the figure, the management table information acquisition unit 5-1 acquires the “work status flag” of the management table 2-4-1 of the management table storage unit 2-4, and confirms the progress of the work. This confirmation process is repeated until the “work status flag” in the management table 2-4-1 becomes “1”, and the work status is monitored. When all “work status flags” become “1”, the contents of the management table 2-4-1 are read.

  The transcription extracting unit 5-2 “transcribes” sequentially from the divided audio file whose “Index” is “1” based on the contents of the management table 2-4-1 read by the management table information acquisition unit 5-1. The text of the item is extracted and concatenated to the last record, text data for the original audio file before division is generated and stored in the text data storage unit 6.

  Next, FIG. 10 is a block diagram showing the configuration of the client terminals 8-1 to 8-n according to the present embodiment. In the figure, the receiving unit 9-1 receives “Index”, “file name”, divided audio file, and “recognition result” from the transcription control device 4. The information display unit 9-2 displays the received information on the terminal screen so that the received information can be browsed. The audio file reproducing unit 9-3 reproduces the received divided audio file with a speaker or a headset. The work staff confirms the “recognition result” while listening to the divided speech.

  When the transcription input unit 9-4 receives an instruction from the work staff that there is no error in the “recognition result”, the transcription input unit 9-4 copies it as it is as a transcription text. On the other hand, when an instruction that the “recognition result” is incorrect is received from the work staff, the text is corrected by the text input by the work staff and the correct content is set as a transcription text. Repeat for all received audio files. The transmission unit 9-5 transcribes “Index” and the transcription text at any time for the divided audio file for which the transcription work has been completed, and transmits it to the control device 4 by e-mail, for example.

Next, the operation of the above-described embodiment will be described.
FIG. 11 is a flowchart for explaining the operation of the transcription text creation support apparatus according to the present embodiment. First, the voice dividing device 2 reads the voice data from the voice data storage unit 1 and divides the voice data into relatively short voice sections (for example, sentence units) using a silent section during speech (step S1). ). Next, the identifier assigning device 3 performs voice recognition on each divided voice file (step S2), and executes matching on the recognition result using the keyword list in the keyword storage unit 3-2 (step S2). S3).

  Next, the transcription control device 4 determines whether or not a keyword is included in the recognition result of the divided audio file (step S4). If a keyword related to confidential information is included, the transcription control device 4 writes it to a highly reliable staff. In order to request a wake-up, the corresponding divided audio file is transmitted to the client terminal of the highly reliable staff (step S5), and the execution of the transcription work by the staff is instructed (step S6). On the other hand, if the keyword related to confidential information is not included in the recognition result of the divided audio file, the corresponding divided audio file is transmitted to the client terminal of the general staff in order to request the general staff to transcribe (step S7). Instruct the execution of the transcription work by the staff (step S8). In any case, when the transcription work is finished, the work result transmitted from the client terminal is received (step S9), the transcription text is extracted and integrated, and the transcription text data of the original voice data before division. Is stored in the text storage unit 6 (step S10).

  According to the above-described embodiment, the risk of information leakage can be greatly reduced by dividing the voice data and requesting a staff member with high reliability to transcribe a portion including confidential information.

  Also, if you want to keep confidentiality without division, you have no choice but to ask all the voice data including the non-confidential part to a highly reliable staff such as a dedicated staff. Is expensive, so it is expensive. According to the above-described embodiment, it is possible to request voice data without confidential information to a general staff with a low unit price with peace of mind, and to reduce the cost as a whole.

  In the above-described embodiment, a program for realizing the functions of the voice dividing device 2, the identifier assigning device 3, the transcription control device 4, the text extracting device 5, and the like is recorded on a computer-readable recording medium. Functions such as the voice dividing device 2, the identifier assigning device 3, the transcription control device 4, and the text extracting device 5 may be realized by causing a computer device to read and execute a program recorded on a recording medium. Here, the “computer system” may include an OS and hardware such as peripheral devices. Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW system is used. The “computer-readable recording medium” means a flexible disk, a magneto-optical disk, a ROM, a writable nonvolatile memory such as a flash memory, a portable medium such as a CD-ROM, a hard disk built in a computer system, etc. This is a storage device.

  Further, the “computer-readable recording medium” means a volatile memory (for example, DRAM (Dynamic DRAM) in a computer system that becomes a server or a client when a program is transmitted through a network such as the Internet or a communication line such as a telephone line. Random Access Memory)), etc., which hold programs for a certain period of time. The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

It is a block diagram which shows the structure of the transcription text creation assistance apparatus by embodiment of this invention. It is a block diagram which shows the structure of the audio | voice division | segmentation apparatus 2 by this embodiment. It is a conceptual diagram which shows the data structure of the management table 2-4-1 by this embodiment. It is a block diagram which shows the structure of the identifier provision apparatus 3 by this embodiment. It is a conceptual diagram which shows an example of the keyword table 3-2-1 by this embodiment. It is a block diagram which shows the structure of the transcription control apparatus 4 by this embodiment. It is a conceptual diagram which shows the data structural example of the work staff information table 4-2-1 by this embodiment. It is a conceptual diagram for demonstrating the state transition of the management table 2-4-1 by this embodiment. It is a block diagram which shows the structure of the text extraction apparatus 5 by this embodiment. It is a block diagram which shows the structure of the client terminals 8-1 to 8-n by this embodiment. It is a flowchart for demonstrating operation | movement of the transcription text creation assistance apparatus by this embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Audio | voice data memory | storage part 2 Audio | voice division | segmentation apparatus 2-1 Management table production | generation part 2-2 Silent section threshold value calculation part 2-3 Voice division | segmentation part 2-4 Management table memory | storage part 2-4-1 Management table 2-5 Division | segmentation audio | voice file Storage unit 3 Identifier assigning device 3-1 Speech recognition unit 3-2 Keyword storage unit 3-2-1 Keyword table 3-3 Keyword collation unit 4 Transcription control device 4-1 Staff information acquisition unit 4-2 Work staff information storage Section 4-2-1 Work staff information table 4-3 Management table information acquisition section 4-4 Transmission section 4-5 Reception section 5 Text extraction device 5-1 Management table information acquisition section 5-2 Transcription extraction section 6 Text data Storage unit 7 Network 8-1 to 8-n Client terminal 9-1 Reception unit 9-2 Information display unit 9-3 Audio file playback unit 9-4 Causing the input unit 9-5 transmission unit

Claims (5)

A transcription text creation support device for transcription of text data from voice data,
For each transcription worker, reliability storage means for storing a reliability value corresponding to the reliability of each transcription worker;
Divided audio data generating means for dividing the audio data into predetermined sections and generating divided audio data;
Conversion means for performing voice recognition for each divided voice data divided by the divided voice data generating means, and converting it into a plurality of primary text data;
Keyword storage means for storing keywords relating to confidential information;
Determination means for determining whether or not the keyword stored in the keyword storage means is included for each primary text data converted by the conversion means;
Based on the determination result by the determination means, the reliability value of the reliability storage means is referred to, and for each divided voice data divided by the divided voice data generation means, a transcription worker who should request a transcription work is determined. A transcription text creation support apparatus comprising: a selection means for selecting. Transmitting means for transmitting the divided audio data to the terminal of the transcription operator selected by the selecting means;
Receiving means for receiving text data of a transcription work result by the transcription worker;
The transcription text creation support apparatus according to claim 1, further comprising: a storage unit that integrates the text data received by the reception unit and stores it as transcription text data with respect to the original voice data. The selecting means is
With reference to the reliability value of the reliability storage means, if the determination means determines that a keyword related to confidential information is included in the primary text, a transcription operator having a high reliability value is selected. The transcription text according to claim 1 or 2, wherein if it is determined that a keyword related to confidential information is not included in the primary text, a transcription worker having a low reliability value is selected. Creation support device. A transcribed text creation support program that supports a transcription work for transcribing text data from audio data,
On the computer,
Dividing the audio data into predetermined intervals to generate divided audio data;
Performing voice recognition for each divided divided voice data and converting it into a plurality of primary text data;
Determining whether or not a keyword related to confidential information is included for each converted primary text data;
When it is determined that a keyword related to confidential information is included in the primary text, a transcription worker having a high reliability value is selected, and it is determined that a keyword related to confidential information is not included in the primary text. And a step of selecting a transcription worker having a low reliability value. A transcribed text creation support method for supporting a transcription work to transcribe text data from voice data executed on a computer ,
Divided speech data generating means, and generating a divided speech data by dividing the audio data into a predetermined interval,
A conversion means for performing voice recognition for each divided divided voice data and converting the voice data into a plurality of primary text data;
A step of determining whether or not a keyword relating to confidential information is included for each converted primary text data;
If the selection means determines that the primary text includes a keyword related to confidential information , the selecting means selects a transcription worker having a high reliability value, and the primary text includes a keyword related to confidential information. And a step of selecting a transcription worker having a low reliability value if it is determined that there is no confidence value.

Images