JP2008009693A - Transcribing system, its server, and program for server - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a transcribing system capable of reducing the burden on an operator of transcribing operation and the burden on a terminal used by the operator compared to a conventional system. <P>SOLUTION: The transcribing system 10 comprises a plurality of transcribing terminals 60 engaged in transcribing operation for generating character data based on voice data, and a management server 40 communicating with the plurality of transcribing terminals 60. The management server 40 comprises an original voice data dividing means for dividing original voice data which is voice data used as the source of transcribing operation, into a plurality of transmission voice data which are voice data to be transmitted to the transcribing terminals 60, and a distributing transmission means for distributing and transmitting the plurality of transmission voice data generated by the voice data dividing means, to the plurality of transcribing terminals 60. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a transcription system for generating character data based on voice data.

As a conventional transcribe system, audio data is transmitted to each operator in real time or at a timing close to it by streaming delivery, and each operator simultaneously edits an arbitrary portion of the retranslation data, thereby allowing one operator to generate one retranslation data. A digital shorthand system to be generated is known (for example, see Patent Document 1).
JP 2004-77966 A

  However, in the conventional awakening system, it is necessary to protect data by applying write protection in line units between all terminal devices in order to avoid updating the same location by a plurality of operators. There was a problem that the burden was large. Also, since no mechanism has been proposed to protect the voice corresponding to a protected text line, the operator only has to listen to the voice to determine whether or not to wake up. Become. Furthermore, when a plurality of workers start updating the same line at the same time, it is before protection is applied, so that the operations may be duplicated and burden the operator.

  The present invention has been made in order to solve the conventional problems, and provides a transcribe system capable of reducing the burden on the worker of the awakening work and the burden on the terminal used by the worker as compared with the prior art. The purpose is to do.

  The server of the present invention is a server that communicates with a plurality of worker terminals that are used by a plurality of workers engaged in a speech task that generates character data based on voice data, and Generated by the original voice data dividing means, the original voice data dividing means for dividing the original voice data that is the voice data to be divided into a plurality of transmission voice data that is the voice data for transmission to the worker terminal. A distribution transmission unit configured to distribute and transmit the plurality of transmission audio data to the plurality of worker terminals;

  With this configuration, the server according to the present invention transmits the transmission voice data to be woken up by the worker of the listening work among the plurality of transmission voice data to the worker terminal. Thus, the burden on the operator can be reduced as compared with the prior art. Moreover, since the server of this invention does not require the synchronization of the character data between the terminals for workers, the burden of the terminal for workers can be reduced conventionally. In addition, since the server of the present invention can simultaneously share the work of arousing the original voice data among a plurality of workers, it is possible to reduce the time required for the whole voice of the voice data. In addition, since the server of the present invention electronically sends the transmission voice data to the worker terminal, for example, the server administrator stores the transmission voice data in a recording medium and mails it to the worker. Thus, it is possible to reduce the time and cost required for the entire original voice data.

  Further, it is preferable that the original voice data dividing unit of the server of the present invention generates the transmission voice data with a silent part included in the original voice data as a delimiter.

  With this configuration, the server according to the present invention reduces the need for the operator to perform operations such as re-listening and pause, as compared with the configuration in which the audio data for transmission is distributed by streaming. It is possible to reduce the time required for the work of revoking the trusted sound data, and to reduce the time required for the work of revoking the entire original sound data.

  Further, it is preferable that the original voice data dividing unit of the server of the present invention generates the transmission voice data with a break portion of a sentence included in the original voice data as a delimiter.

  With this configuration, the server according to the present invention can reduce the time required for the worker to wake up the transmission voice data compared to a configuration in which the transmission voice data is too large. It is possible to shorten the time required for the work of listening. In addition, the server of the present invention enables the worker of the work to easily hear the contents of the story included in the voice data for transmission compared with the configuration in which the voice data for transmission is too small. It is possible to facilitate the task of awakening the voice data for transmission.

  In addition, it is preferable that the original voice data dividing unit of the server of the present invention generates the transmission voice data with a speaker switching portion included in the original voice data as a delimiter.

  With this configuration, the server according to the present invention can reduce the time required for the worker to wake up the transmission voice data compared to a configuration in which the transmission voice data is too large. It is possible to shorten the time required for the work of listening. In addition, the server of the present invention enables the worker of the work to easily hear the contents of the story included in the voice data for transmission compared with the configuration in which the voice data for transmission is too small. It is possible to facilitate the task of awakening the voice data for transmission.

  In addition, it is preferable that the original voice data dividing unit of the server of the present invention generates the transmission voice data with a topic switching portion included in the original voice data as a delimiter.

  With this configuration, the server according to the present invention can reduce the time required for the worker to wake up the transmission voice data compared to a configuration in which the transmission voice data is too large. It is possible to shorten the time required for the work of listening. In addition, the server of the present invention enables the worker of the work to easily hear the contents of the story included in the voice data for transmission compared with the configuration in which the voice data for transmission is too small. It is possible to facilitate the task of awakening the voice data for transmission.

  Moreover, it is preferable that the said distribution transmission means of the server of this invention selects the said operator terminal which transmits the said audio data for transmission according to the speaker of the said audio data for transmission.

  With this configuration, the server according to the present invention can be specialized for a specific speaker for each worker, so that the efficiency of the wake-up work by the worker can be improved. In addition, since the server of the present invention can be specialized for a specific speaker for each worker terminal, when the worker terminal performs speech recognition, the accuracy of speech recognition depends on the voice quality or topic of the speaker. Can be improved.

  The server of the present invention further includes character data generation means for generating the character data based on the voice data by voice recognition, and the distribution transmission means transmits the transmission voice data to the worker terminal. It is preferable that the character data corresponding to is transmitted to the worker terminal.

  With this configuration, the server of the present invention generates a large amount of character data by voice recognition, so that the worker can listen to the transmission voice data as compared with a configuration in which the transmission voice data is aroused from the beginning. The time and labor required for the wake-up work can be reduced. Therefore, the server of the present invention can reduce the time and labor cost required for the whole original voice data to be aroused.

  The distribution transmission unit of the server of the present invention transmits the transmission voice data to the worker terminal that has made the request when the request for the transmission voice data is received from the worker terminal. It is preferable to do.

  With this configuration, the server of the present invention can transmit the audio data for transmission to the worker terminal according to the progress status of the audible work by the worker, so what is the progress status of the wake-up work by the worker? Irrespective of the configuration in which the transmission voice data is transmitted to the worker terminal regardless of the above, it is possible to streamline the entire original voice data and reduce the required time.

  The server of the present invention further includes a transmission voice data selection unit that selects the transmission voice data transmitted by the distribution transmission unit, and the transmission voice data selection unit includes a topic of the transmission voice data. It is preferable to select the transmission voice data based on the operator.

  With this configuration, the server according to the present invention can transmit the transmission voice data suitable for the worker to the worker terminal, and therefore the transmission voice data is transmitted to the worker terminal regardless of the topic. Compared with, it is possible to shorten the time required for the operator to wake up. Therefore, the server of the present invention can shorten the time required for the work of rehearsing the entire original voice data.

  The server of the present invention further includes a transmission voice data selection unit that selects the transmission voice data transmitted by the distribution transmission unit, and the transmission voice data selection unit is configured such that the worker terminal has a voice. When generating the character data based on the voice data by recognition, the voice data for transmission is determined based on the topic of the voice data for transmission and the topic targeted by the voice recognition dictionary used by the voice recognition. It is preferable to select.

  With this configuration, the server of the present invention can transmit the transmission voice data suitable for the worker terminal to the worker terminal, and therefore transmits the transmission voice data to the worker terminal regardless of the topic. Compared with the structure to perform, it can shorten the time which an operator requires for arousing work. Therefore, the server of the present invention can shorten the time required for the work of rehearsing the entire original voice data.

  Moreover, it is preferable that the said distribution transmission means of the server of this invention encrypts and transmits the said audio | voice data for transmission to the said terminal for workers.

  With this configuration, the server of the present invention communicates with the worker's terminal via the public network while protecting the secret even when the confidential information such as confidential information and personal information is included in the original voice data. be able to.

  Further, the server program of the present invention is a server program for operating a server that communicates with a plurality of worker terminals used by a plurality of workers engaged in a transcription work that generates character data based on voice data. And original voice data dividing means for dividing the original voice data which is the voice data which is the source of the audible work into a plurality of transmission voice data which is voice data for transmission to the worker terminal, The server is made to function as a distribution transmission unit that distributes and transmits the plurality of transmission audio data generated by the original audio data division unit to the plurality of worker terminals.

  With this configuration, the server program of the present invention transmits the voice data for transmission to be woken up by the worker of the listening work among the plurality of voice data for transmission to the worker terminal. It is more efficient and the burden on the operator can be reduced than before. Moreover, since the server program of the present invention does not require the synchronization of character data between the worker terminals, the burden on the worker terminal can be reduced as compared with the prior art. In addition, the server program of the present invention can share the work of rehearsing the original voice data to a plurality of workers at the same time, so that the time required for the work of rehearsing the whole original sound data can be reduced. In addition, since the server program of the present invention electronically sends the transmission voice data to the worker terminal, the original voice is compared with the case where the transmission voice data is stored in the recording medium and mailed to the worker. It is possible to reduce the time and cost required for the whole data rehearsal work.

  In addition, the speech system according to the present invention includes a plurality of worker terminals used by a plurality of workers engaged in a speech task that generates character data based on voice data, and communicates with the plurality of worker terminals. And a server for performing a speech process, wherein the server is a plurality of voice data for transmission to the worker terminal of original voice data that is voice data that is a source of the work to be heard Original voice data dividing means for dividing the voice data for transmission, and distribution transmitting means for distributing the plurality of transmission voice data generated by the original voice data dividing means to the plurality of worker terminals for transmission. It is characterized by having.

  With this configuration, the speech system according to the present invention transmits the voice data for transmission to be heard by the worker of the speech work among the plurality of voice data for transmission to the worker terminal. It is more efficient and the burden on the operator can be reduced than before. Moreover, since the transcription system of this invention does not require the synchronization of the character data in worker terminals, the burden of a worker terminal can be reduced conventionally. In addition, since the audible system of the present invention can share the utterance work of the original voice data to a plurality of workers at the same time, the time required for the wakeup work of the entire original voice data can be shortened. In addition, since the transmission system of the present invention electronically transmits the transmission voice data to the worker terminal, the original voice is compared with the case where the transmission voice data is stored in the recording medium and mailed to the worker. It is possible to reduce the time and cost required for the whole data rehearsal work.

  Further, the worker terminal of the speech system of the present invention, the voice output means for outputting the voice based on the voice data, the character data editing means for editing the character data according to the instructions of the worker, It is preferable to include character data transmission means for transmitting the character data to the server.

  With this configuration, the speech system according to the present invention electronically sends the character data aroused by the worker from the worker terminal to the server. For example, the worker stores the character data in the recording medium and stores the character data on the server. Compared with the case of mailing to an administrator, the time and cost required for the work of revoking the entire original voice data can be reduced.

  In addition, the distribution transmission means of the speech system according to the present invention includes delimiter information including at least one of a silent part, a sentence break part, a speaker switching part, and a topic switching part included in the original voice data. The worker terminal is transmitted to the worker terminal together with the transmission voice data, and the worker terminal is a plurality of voice data for outputting the voice by the voice output means based on the delimiter information. It is preferable to include transmission audio data dividing means for dividing the output audio data into the output audio data.

  With this configuration, the speech system according to the present invention can output an amount of sound that remains in the worker's short-term memory, and thus outputs a large amount of sound continuously so as not to remain in the worker's short-term memory. Compared with the structure to perform, it is possible to improve the efficiency of the worker's awakening work. Therefore, the rehearsal system of the present invention can reduce the time required for rehearsing the entire original voice data.

  Further, it is preferable that the character data editing means of the speech system of the present invention edits the character data for each output audio data.

  With this configuration, the speech system according to the present invention can cause the worker to edit the character data at the location corresponding to the voice being output, so that the efficiency of the speech work by the worker can be improved. Therefore, the rehearsal system of the present invention can reduce the time required for rehearsing the entire original voice data.

  Moreover, it is preferable that the said operator terminal of the transcription system of this invention is provided with the character data production | generation means which produces | generates the said character data based on the said audio | voice data by audio | voice recognition.

  With this configuration, the speech system according to the present invention generates a large amount of character data by speech recognition, so that the worker can transmit the voice data for transmission compared to the configuration in which the worker transmits the voice data for transmission from the beginning. It is possible to reduce the time and labor required for the task of awakening. Therefore, the speech system according to the present invention can reduce the time and labor cost required for the entire speech processing of the original voice data.

  Further, the distribution transmission means of the transcribe system of the present invention encrypts and transmits the transmission voice data to the worker terminal, and the character data transmission means encrypts the character data to the server. Are preferably transmitted.

  With this configuration, the speech system according to the present invention allows the public system to keep the secret between the server and the worker terminal even when the original voice data includes confidential information such as confidential information and personal information. Communication can be performed via a network.

  According to the present invention, it is possible to provide a wake-up system that can reduce the burden on the worker of the wake-up work and the load on the terminal used by the worker.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

  First, the configuration of the speech system according to the present embodiment will be described.

  As shown in FIG. 1, a speech system 10 according to the present embodiment is installed in a place such as a public parliament or a company meeting room, and a recording device 20 that records the conference, and the recording device 20 records the recording. A management server 40 serving as a server for managing a speech task that generates character data based on speech, and a plurality of speech terminals 60 serving as a plurality of worker terminals used by a plurality of workers involved in the speech task. And. The recording device 20, the management server 40, and the listening terminal 60 are connected to a public network 11 such as the Internet.

  As shown in FIG. 2, the management server 40 is a computer and is operated by a server program. The server program includes a communication means 41 for communicating with the recording device 20 (see FIG. 1) and the audible terminal 60 (see FIG. 1), a storage means 42 for storing various data, and a wake-up work. Original voice data 80 (see FIG. 3) that is the original voice data is awakened and divided into a plurality of transmission voice data 81 (see FIG. 3) that is voice data for transmission to the terminal 60. The management server 40 is caused to function as the dividing means 43 and the transmission voice data selection means 44 for selecting the transmission voice data 81 transmitted by the communication means 41.

  The communication means 41 distributes a plurality of transmission voice data 81 generated by the original voice data dividing means 43 to a plurality of audible terminals 60 via the public network 11 and transmits them. It is supposed to function as well. Note that when the communication means 41 hears a request for the transmission voice data 81 and receives the request from the terminal 60, the communication means 41 transmits the transmission voice data 81 to the requesting call terminal 60.

  Further, the storage means 42 is an order table (see FIG. 4) showing the correspondence between the order from the beginning of the original voice data 80 and the listening terminal 60 as the transmission destination for the voice data 81 for transmission. Is to be remembered. In the “transmission destination” item of the order table, a value of “not yet” is stored for the transmission voice data 81 that is not yet transmitted to the terminal 60 by the management server 40, and the management server 40 awakens the terminal 60. For the transmission voice data 81 transmitted to, the name of the transmission terminal 60 is stored, and for the transmission voice data 81 for which the management server 40 has received the corresponding character data from the terminal 60 Is stored.

  Further, the original voice data dividing unit 43 determines a sentence break included in the original voice data 80 by using the voice recognition function and the natural language processing function, and generates the transmission voice data 81 with the sentence break as a delimiter. It is like that. That is, the original voice data dividing unit 43 generates the transmission voice data 81 in sentence units.

  Further, the transmission voice data selection means 44 selects the transmission voice data 81 based on the topic of the transmission voice data 81 and the topic targeted by the voice recognition dictionary used for voice recognition in the listening terminal 60. It is supposed to be.

  As shown in FIG. 5, the awakening terminal 60 is a computer and is operated by a terminal program. The terminal program outputs a communication means 61 for communicating with the management server 40 (see FIG. 1), a storage means 62 for storing various data, and audio data for transmission 81 (see FIG. 6). Transmission voice data dividing means 63 for dividing the output voice data 82 into a plurality of output voice data 82 (see FIG. 6), character data generating means 64 for generating character data by voice recognition, and output voice data The speech terminal 60 is made to function as a voice output means 65 that outputs voice based on 82 and a character data editing means 66 that edits character data in accordance with instructions from the operator.

  The communication means 61 transmits character data to the management server 40, and functions as character data transmission means.

  Further, the transmission voice data dividing means 63 determines the break portion of the phrase included in the transmission voice data 81 by the voice recognition function and the natural language processing function, and outputs the output voice data 82 by using the break portion of the phrase as a delimiter. It is designed to generate. That is, the transmission audio data dividing means 63 generates the output audio data 82 in units of phrases.

  The character data editing unit 66 edits the character data for each output audio data 82.

  Next, the operation of the speech system 10 will be described.

  First, the operation of the recording device 20 will be described.

  When the recording of the conference is instructed by the administrator of the recording device 20, the recording device 20 performs the recording of the conference and records the original voice data 80, while sending the original voice data 80 to the management server 40 via the public network 11. Is sent in real time.

  Next, the operation of the management server 40 will be described.

  When the management server 40 starts to receive the original voice data 80 from the recording device 20 via the public network 11 by the communication means 41, the management server 40 starts to store the original voice data 80 by the storage means 42 as shown in FIG. 7 (S101). ).

  Next, the management server 40 starts to divide the original voice data 80 stored in the storage means 42 into the transmission voice data 81 as shown in FIG. 3 by the original voice data dividing means 43 (S102). Here, the management server 40 generates the transmission voice data 81 by using the breaks of the sentence included in the original voice data 80 as a delimiter. That is, the transmission voice data 81 is sentence-by-sentence voice data.

  Note that, when the transmission server 81 generates the transmission voice data 81, the management server 40 adds an item for the generated transmission voice data 81 to the order table, and stores a value of “not yet” in the “transmission destination” item.

  Then, the management server 40 starts transmitting the plurality of transmission voice data 81 by the communication means 41 to the plurality of listening terminals 60 via the public network 11 (S103). In other words, when the management server 40 subsequently asks for the transmission voice data 81 via the public network 11 and receives the request from the terminal 60 by the communication means 41, the management server 40 connects the public network 11 to the requesting talk terminal 60. Via the communication means 41, transmission voice data 81 is transmitted. However, the management server 40 does not respond to the request from the awakening terminal 60 whose name is stored in the “transmission destination” item of the order table.

  When the management server 40 listens to the transmission voice data 81 and transmits it to the terminal 60, the name of the transmission destination voice terminal 60 is stored in the “transmission destination” item of the order table for the transmitted transmission voice data 81. To do.

  The transmission voice data 81 transmitted by the management server 40 is selected by the transmission voice data selection means 44. In other words, the management server 40 transmits the transmission voice based on the topic of the transmission voice data 81 and the topic targeted by the voice recognition dictionary used by the listening terminal 60 that has requested the transmission voice data 81. The transmission voice data 81 to be transmitted to the listening terminal 60 that has requested the data 81 is selected. For example, when the topic targeted by the speech recognition dictionary used by the listening terminal 60 that has requested the transmission voice data 81 is related to medical care, the management server 40 has requested the transmission voice data 81. As the transmission voice data 81 to be transmitted to the listening terminal 60, the transmission voice data 81 whose topic is related to medical care is selected. The topic of the transmission voice data 81 may be registered in advance based on the agenda of the meeting, or may be extracted from the original voice data 80 or the transmission voice data 81 by the management server 40 by voice recognition. The topic targeted by the speech recognition dictionary used by the awakening terminal 60 that requested the transmission voice data 81 may be registered in the management server 40 in advance, or listened together with the request for the transmission voice data 81. The management server 40 may be notified from the wake-up terminal 60.

  Then, when the management server 40 receives the character data from the terminal 60, which is awakened via the public network 11, by the communication means 41, the management server 40 starts storing the received character data while arranging the received character data (S104). In other words, when the management server 40 subsequently wakes up character data and receives it from the terminal 60 via the public network 11, the management server 40 itself sends to the wake-up terminal 60 which is the transmission source of the received character data. For the trusted voice data 81, the order in the original voice data 80 is obtained based on the order table, and character data is stored while being arranged in the obtained order. That is, the character data received by the management server 40 is arranged in the same order as the original transmission voice data 81 and finally becomes character data corresponding to the original voice data 80 as a whole.

  When the management server 40 stores the character data, the “transmission destination” in the order table for the transmission voice data 81 transmitted immediately before by the management server 40 itself to the listening terminal 60 that is the transmission source of the stored character data. Stores the value “done” in the item.

  Next, the operation of the listening terminal 60 will be described.

  When the operator instructs the capture of the transmission voice data 81, the communication terminal 61 requests the transmission voice data 81 from the communication server 61 to the management server 40 via the public network 11, as shown in FIG. (S121).

  Then, the listening terminal 60 determines whether or not the transmission voice data 81 is transmitted from the management server 40 until it is determined that the transmission voice data 81 is transmitted from the management server 40 via the public network 11 ( S122).

  When the listening terminal 60 determines in S122 that the transmission voice data 81 has been transmitted from the management server 40, the communication terminal 61 receives the transmission voice data 81 transmitted from the management server 40 (S123), and the storage means. 62 (S124).

  Next, the listening terminal 60 divides the transmission audio data 81 stored in S124 into phrase units as output audio data 82 as shown in FIG. 6 by the transmission audio data dividing means 63 (S125), and stores it in S124. Based on the transmitted voice data 81, the character data generating means 64 performs voice recognition to generate character data for each output voice data 82 (S126).

  Then, the awakening terminal 60 determines whether or not the operator has instructed completion of editing all the character data generated in S126 (S127).

  If it is determined in S127 that the completion of editing of all the character data is not instructed by the operator, the listening terminal 60 determines whether any of the output audio data 82 divided in S125 is specified by the operator. Is determined (S128).

  If it is determined in S128 that no voice data 82 for output has been designated by the operator, the awakening terminal 60 executes the process of S127 again.

  When it is determined in S128 that any of the output audio data 82 has been designated by the operator, the awakening terminal 60 outputs the audio by the audio output means 65 based on the output audio data 82 designated by the operator in S128. (S129) After the character data corresponding to the output voice data 82 designated by the operator in S128 is edited by the character data editing means 66 based on the operator's instruction (S130), the process of S127 is executed again. To do. Therefore, the operator edits the character data through an input device such as a keyboard (not shown) of the speech terminal 60 while listening to the voice from a speech output device such as a speaker (not shown) of the speech terminal 60. The work can be executed for each output audio data 82, that is, for each phrase.

  When it is determined in S127 that the operator has instructed completion of editing all the character data, the transcribe terminal 60 transmits a series of character data to the management server 40 via the public network 11 by the communication means 61 (S131). .

  As described above, the awakening system 10 awakens and transmits to the terminal 60 the transmission audio data 81 to be awakened by the worker of the awakening work among the plurality of transmission sound data 81. The work can be made more efficient than before, and the burden on the operator can be reduced more than before.

  In addition, since the speech system 10 does not require synchronization of character data between the speech terminals 60, the burden on the speech terminals 60 can be reduced as compared with the conventional case.

  In addition, since the audible system 10 can simultaneously share the audible work of the original voice data 80 to a plurality of workers, the time required for the wakeup work of the entire original voice data 80 can be reduced.

  In addition, since the audible system 10 audibly transmits the transmission voice data 81 and electronically sends it to the terminal 60, the original voice is compared with the case where the transmission voice data 81 is stored in a recording medium and mailed to the operator. It is possible to reduce the time and cost required for the entire data 80 to be transcribed.

  Then, since the audible system 10 can shorten the time required for the audible work of the entire original voice data 80, for example, a minutes describing the contents of the morning meeting can be prepared in the afternoon.

  In addition, since the awakening system 10 awakens the transmission voice data 81 and sends it electronically to the terminal 60, it provides a working opportunity to a person who is motivated to work but is restricted by the conditions of working from home, for example, a single mother. Can do.

  In addition, the transcribe system 10 transcribes character data aroused by the worker and electronically sends it from the terminal 60 to the management server 40. For example, the worker stores the character data in a recording medium and stores the character data in the management server 40. Compared with the case of mailing to the administrator, the time and cost required for the work of awakening the entire original voice data 80 can be reduced.

  In addition, since the management server 40 generates the transmission voice data 81 with the sentence break included in the original voice data 80 as a delimiter, compared to the configuration in which the transmission voice data 81 is too large, It is possible to reduce the time required for the operator to wake up the transmission voice data 81, and to reduce the time required to wake up the entire original voice data 80. In addition, since the management server 40 is configured to generate the transmission voice data 81 with the sentence break included in the original voice data 80 as a delimiter, compared to the configuration in which the transmission voice data 81 is too small, The content of the story contained in the transmission voice data 81 can be aroused to make the worker of the work understand easily, and the worker can easily make the transmission voice data 81 awake.

  Note that the original voice data dividing unit 43 of the management server 40 may generate the transmission voice data 81 with a portion other than the break portion of the sentence included in the original voice data 80 as a delimiter. It is preferable that the transmission voice data 81 is small enough to promptly determine whether or not the worker can wake up and accept the work.

  For example, the original voice data dividing unit 43 determines a silent part such as a breathing included in the original voice data 80 by determining a silent part when a state of a predetermined volume or lower continues for a predetermined number of seconds or more. The transmission audio data 81 may be generated with the silent part as a break. In this case, the management server 40 reduces the need for the operator to perform operations such as re-listening and pause as compared with the configuration in which the transmission audio data 81 is distributed by streaming. It is possible to reduce the time required for the speech work of the voice data 81, and to shorten the time required for the work of rehearsing the original voice data 80 as a whole.

  Further, the original voice data dividing means 43 determines a speaker switching portion included in the original voice data 80 by a speaker identification function using a voiceprint or the like, and sets the transmission voice data 81 by using the speaker switching portion as a delimiter. It may be generated. Further, the original voice data dividing unit 43 determines a topic switching portion included in the original voice data 80 by the voice recognition function and the natural language processing function, and generates transmission voice data 81 with the topic switching portion as a delimiter. It may be like this. In these cases, the management server 40 can obtain the same effect as that of the configuration in which the transmission voice data 81 is generated with the sentence break included in the original voice data 80 as a break.

  The original voice data dividing unit 43 divides the original voice data 80 by separating at least one of a silent part, a sentence break part, and a speaker switching part included in the original voice data 80, and then for a predetermined time. For example, the audio data 81 for transmission may be generated by recombining three minutes. In this case, the management server 40 can make the time required for the worker to hear the transmission voice data 81 substantially uniform.

  Further, the communication means 41 of the management server 40 transmits the transmission voice data 81 when the original voice data division means 43 generates the transmission voice data 81 with the speaker switching part included in the original voice data 80 as a delimiter. It is also possible to select the listening terminal 60 to be selected according to the speaker of the transmission voice data 81. In this case, since the management server 40 can be specialized for a specific speaker for each listening terminal 60, it is possible to improve the accuracy of speech recognition by the terminal 60 based on the voice quality and topic of the speaker. In addition, since the management server 40 can be specialized for a specific speaker for each worker, the efficiency of the awakening work by the worker can be improved.

  In addition, when the management server 40 hears a request for the transmission voice data 81 and receives it from the terminal 60, the management server 40 transmits the transmission voice data 81 to the requesting voice terminal 60. The transmission voice data 81 can be transmitted to the listening terminal 60 in accordance with the progress of the listening work by the person. Therefore, the management server 40 is more efficient in performing the wake-up work of the entire original sound data 80 as compared to the configuration in which the sound data 81 for transmission is auded and transmitted to the terminal 60 regardless of the progress of the wake-up work by the worker. To reduce the required time.

  Further, the management server 40 selects the transmission voice data 81 based on the topic of the transmission voice data 81 and the topic targeted by the voice recognition dictionary used by the voice recognition in the listening terminal 60. Therefore, the transmission voice data 81 suitable for the listening terminal 60 can be heard and transmitted to the terminal 60. Therefore, the management server 40 can shorten the time required for the operator to wake up the work compared to the configuration in which the voice data 81 for transmission is audible and transmitted to the terminal 60 regardless of the topic. It is possible to reduce the time required for the entire 80 rehearsal operation.

  The transmission voice data selection unit 44 of the management server 40 selects the transmission voice data 81 by a method other than the method based on the topic of the transmission voice data 81 and the topic targeted by the voice recognition dictionary. It may be.

  For example, the transmission voice data selection unit 44 may select the transmission voice data 81 based on the topic of the transmission voice data 81 and the worker in the listening terminal 60. In this case, for example, the management server 40 transmits the transmission voice data 81 on the topic related to the law to the worker's awakening terminal 60 who is good at the topic related to the law. Can be transmitted to the terminal 60, so that the time required for the operator to wake up can be shortened compared to a configuration in which the transmission voice data 81 is audible and transmitted to the terminal 60 regardless of the topic. Can do. Therefore, the management server 40 can shorten the time required for the task of awakening the entire original voice data 80.

  In addition, since the awakening system 10 generates almost character data by voice recognition, the worker can listen to the transmission voice data 81 in comparison with a configuration in which the transmission voice data 81 is aroused from the beginning. The time and labor required for the wake-up work can be reduced. Therefore, the rehearsal system 10 can reduce the time and labor cost required for rehearsing the entire original voice data 80.

  In the speech system 10, the speech terminal 60 does not generate character data by voice recognition, but the management server 40 generates character data by voice recognition based on the original voice data 80, and transmission voice data 81. The management server 40 may awaken and transmit the character data corresponding to to the terminal 60 together with the transmission voice data 81.

  Of course, both the management server 40 and the listening terminal 60 may not generate character data by voice recognition.

  In addition, since the listening terminal 60 divides the transmission voice data 81 into the output voice data 82, it is possible to output an amount of voice that remains in the worker's short-term memory. Therefore, the rehearsal system 10 improves the efficiency of the requisition work by the operator as compared with the configuration in which the terminal 60 continuously outputs a large amount of sound that does not remain in the worker's short-term memory. Thus, it is possible to shorten the time required for the entire original voice data 80 to be transcribed.

  The output audio data 82 is in phrase units in the present embodiment, but may be other than phrase units. For example, the awakening terminal 60 manages the delimiter information including at least one of a silent part, a sentence break part, a speaker switching part, and a topic switching part included in the transmission voice data 81 together with the transmission voice data 81. Based on the delimiter information, the output audio data 82 is separated from at least one of a silence portion, a sentence break portion, a speaker switching portion, and a topic switching portion received from the server 40 based on the delimiter information. It may be generated. When the transmission voice data 81 is generated with a speaker switching portion in the original voice data 80 as a delimiter, the awakening terminal 60 uses, for example, a sentence break included in the transmission voice data 81 as a delimiter. The output audio data 82 may be generated.

  Of course, the awakening terminal 60 may be configured to continuously output the voice of the entire transmission voice data 81 without dividing the transmission voice data 81.

  In addition, since the awakening system 10 edits the character data for each output voice data 82, the operator can edit the character data at the location corresponding to the voice being output. Therefore, the rehearsal system 10 can improve the efficiency of the rehearsal work by the operator, and can reduce the time required for the rehearsal work of the entire original voice data 80.

  Of course, the awakening terminal 60 may not be able to edit the character data for each output audio data 82.

  The recording device 20 may encrypt the original voice data 80 when transmitting the original voice data 80 to the management server 40. The management server 40 may encrypt the transmission voice data 81 when the transmission voice data 81 is heard and transmitted to the terminal 60. The audible terminal 60 may encrypt the character data when transmitting the character data evoked by the worker to the management server 40. When the listening system 10 performs encrypted communication among the recording device 20, the management server 40, and the listening terminal 60, confidential information such as confidential information and personal information is included in the original voice data 80, for example. Even in such a case, communication can be performed via the public network 11 while protecting the secret among the recording device 20, the management server 40, and the listening terminal 60.

  Further, the original voice data 80 is stored in the management server 40 from the recording device 20 via the public network 11, but may be stored in the management server 40 by other methods. . For example, the original audio data 80 is stored in a recording medium by the administrator of the recording device 20 and mailed to the administrator of the management server 40, and is stored in the management server 40 by the administrator of the management server 40. Also good.

  Note that the awakening system 10 can also be applied to various businesses other than the minutes creation business. For example, the speech system 10 inputs voice to a mobile phone, generates original voice data 80 in the management server 40, shares the generated original voice data 80 with a plurality of voice terminals 60, wakes up, and wakes up. The character data transmitted from the management server 40 to the mobile phone and the character data transmitted from the management server 40 to the mobile phone can be used as a mail text by the mobile phone.

  Further, the terminal program and the server program may be stored and distributed in a recording medium, or may be distributed via a network.

The block diagram which shows the structure of the transcription system which concerns on one embodiment of this invention The block diagram which shows the function of the management server shown in FIG. The figure which shows the original audio | voice data handled in the speech system shown in FIG. 1, Comprising: The figure which shows the state divided | segmented into the audio | voice data for transmission The figure which shows the order table memorize | stored in the management server shown in FIG. Block diagram showing the function of the listening terminal shown in FIG. FIG. 4 is a diagram showing the audio data for transmission shown in FIG. 3, and shows a state where the audio data is divided into output audio data The flowchart which shows operation | movement of the management server shown in FIG. Flowchart showing the operation of the listening terminal shown in FIG.

Explanation of symbols

10 Transcription System 11 Public Network 20 Recording Device 40 Management Server (Server)
41 Communication means (distribution transmission means)
42 Storage means 43 Original voice data division means 44 Transmission voice data selection means 60 Talking terminal (worker terminal)
61 Communication means (character data transmission means)
62 Storage means 63 Transmission voice data division means 64 Character data generation means 65 Voice output means 66 Character data editing means 80 Original voice data 81 Transmission voice data 82 Output voice data

Claims (18)

A server that communicates with a plurality of worker terminals used by a plurality of workers engaged in a speech task that generates character data based on voice data,
Original voice data dividing means for dividing original voice data, which is voice data that is a source of the awakening work, into a plurality of transmission voice data that is voice data for transmission to the worker terminal; and the original voice data A server comprising: a distribution transmission unit that distributes and transmits the plurality of transmission audio data generated by the division unit to the plurality of worker terminals.   2. The server according to claim 1, wherein the original voice data dividing unit generates the transmission voice data with a silent part included in the original voice data as a delimiter. 3.   2. The server according to claim 1, wherein the original voice data dividing unit generates the transmission voice data using a break portion of a sentence included in the original voice data as a delimiter.   2. The server according to claim 1, wherein the original voice data dividing unit generates the transmission voice data with a speaker switching portion included in the original voice data as a delimiter. 3.   2. The server according to claim 1, wherein the original voice data dividing unit generates the transmission voice data with a topic switching portion included in the original voice data as a delimiter. 3.   The server according to claim 3, wherein the distribution transmission unit selects the worker terminal that transmits the transmission voice data according to a speaker of the transmission voice data. Character data generating means for generating the character data based on the voice data by voice recognition,
The server according to claim 1, wherein the distribution transmission unit transmits the character data corresponding to the transmission voice data to be transmitted to the worker terminal to the worker terminal.   The distribution transmission unit transmits the transmission voice data to the worker terminal that has made the request when the request for the transmission voice data is received from the worker terminal. Item 4. The server according to item 1. Voice data selection means for transmission for selecting the voice data for transmission transmitted by the distribution transmission means,
The server according to claim 8, wherein the transmission voice data selection unit selects the transmission voice data based on a topic of the transmission voice data and the worker. Voice data selection means for transmission for selecting the voice data for transmission transmitted by the distribution transmission means,
The transmission voice data selection means, when the worker terminal generates the character data based on the voice data by voice recognition, the topic of the voice data for transmission and the voice recognition used by the voice recognition 9. The server according to claim 8, wherein the transmission voice data is selected based on a topic targeted by a dictionary.   The server according to claim 1, wherein the distribution transmission unit encrypts and transmits the transmission voice data to the worker terminal. A server program for operating a server that communicates with a plurality of worker terminals used by a plurality of workers engaged in a speech task that generates character data based on voice data,
Original voice data dividing means for dividing original voice data, which is voice data that is a source of the awakening work, into a plurality of transmission voice data that is voice data for transmission to the worker terminal; and the original voice data A server program that causes a server to function as a distribution transmission unit that distributes and transmits a plurality of transmission audio data generated by a division unit to the plurality of worker terminals. A speech system comprising a plurality of worker terminals used by a plurality of workers engaged in a speech task that generates character data based on voice data, and a server that communicates with the plurality of worker terminals Because
The server includes original voice data dividing means for dividing original voice data, which is voice data that is a source of the listening work, into a plurality of transmission voice data that is voice data for transmission to the worker terminal; 2. A hearing system comprising: a distribution transmission unit that distributes and transmits a plurality of the transmission audio data generated by the original audio data division unit to the plurality of worker terminals.   The worker terminal transmits voice data to the server, voice output means for outputting voice based on the voice data, character data editing means for editing the character data in accordance with an instruction from the worker, and the server. The speech system according to claim 13, further comprising character data transmission means. The distribution transmission means includes at least one type of delimiter information included in the original voice data, including a silent part, a sentence break part, a speaker switching part, and a topic switching part, together with the transmission voice data. To the device for
The worker terminal includes transmission voice data dividing means for dividing the transmission voice data into a plurality of output voice data which are voice data for outputting the voice by the voice output means based on the delimiter information. 15. The speech system according to claim 14, further comprising:   16. The system according to claim 15, wherein the character data editing unit edits the character data for each output voice data.   The speech system according to claim 14, wherein the worker terminal includes character data generation means for generating the character data based on the voice data by voice recognition. The distribution transmission means encrypts and transmits the transmission voice data to the worker terminal,
15. The system according to claim 14, wherein the character data transmission means encrypts the character data and transmits it to the server.

Images