WO2018227761A1 - Correction device for recorded and broadcasted data for teaching - Google Patents

Abstract

The present invention provides a correction device for recorded and broadcasted data for teaching. Voice recording equipment is used to convert voice signals generated in a process of network teaching or online meeting into original voice data having time stamps; a voice recognition model is used to recognize and convert the original voice data into original text data; the original text data is proofread to substitute old text content needed to be corrected with new text content, thereby implementing correction of the original text data to form corrected text data; the time stamps are used for positioning; standard voice data of the new text content is used to substitute the corresponding voice data slots of the old text content to form corrected voice data. The use of the device of the present invention can correct recorded and broadcasted data for teaching, thus solving confusion and misleading to a user in recording and broadcasting systems for teaching resulted from speaking errors and omissions and nonstandard expressions existing in classroom teaching.

Description

Teaching recording and broadcasting data correction device Technical field

The invention relates to a network teaching recording and broadcasting technology, which can be used for recording and playing a teaching activity or a conference process based on network teaching or online conference, and particularly relates to a device capable of correcting recorded teaching voice data.

Background technique

In recent years, as traditional teaching models have become less and less able to meet the needs of users for new teaching methods such as multimedia, informationization, and easy playback, with the rapid development and popularization of Internet technologies, especially mobile Internet technologies, various online teaching The recording and broadcasting system is booming. In the network teaching, the teaching process is recorded through the classroom recording, and the teaching resources can be shared on the Internet. The user can use the terminal to access these teaching resources online, which can meet the needs of the user for remote learning and review.

Early technologies in teaching recording and broadcasting, such as CN101141271A (publication date March 12, 2008), disclosed a recording and broadcasting system for network teaching, including: recorder, processor, first network, second network, server, database And three clients. Among them, the recorder mainly includes a camera and a wireless digital microphone to record video information and voice data of the courseware. The first network transmits the courseware information to the server. The server is used on the one hand to further process the courseware information, to generate courseware data, and on the other hand to search and call the courseware data in the database, and then convert the courseware data back to the courseware information. The database is used to store the courseware data. The second network is used to connect the client to the server. The client is used to facilitate the user to query courseware information and invoke courseware information. Said The patent application discloses a relatively typical streaming media format recording course technology. Now, its main disadvantage is that the files formed after recording are relatively large, the uploading and downloading speed is slow, and the required storage space is large.

Recent technologies in teaching and recording, such as CN105306861A (publication date February 3, 2016), disclose an effective classroom teaching and recording method and system. In the process of online teaching or online meeting, multimedia whiteboard can be realized for users. The functional voice, speech/speech voice, communication with other users, and/or coaching, etc., are recorded to form different data streams, and a unified time stamp for various data streams is generated by the network teaching recording system. Marking, instead of recording the entire event in a streaming media format, the network user can conveniently download various data streams to be played from the cloud server or the LAN server through the network anytime and anywhere, and obtain the data stream after the customer of the user terminal. The end obtains the data stream according to the time stamp reproduction, and the organic combination plays out to display to the user, thereby completing the on-demand browsing. The patent application discloses a classroom recording and recording method for separately storing and recording classroom teaching data in three data stream formats according to time stamps.

With the increasing pursuit of the quality of recorded courses, more and more teaching and recording systems use speech recognition technology, usually need to convert the speech into text, display it as a subtitle on the screen or save it as a text format. In the prior art, there are a few patent applications for speech recognition, especially converting speech to text or converting text to speech, such as:

CN101354748A (Publication Date January 28, 2009) discloses a character recognition device including an image pickup device, a character recognition device, a voice conversion device, and a voice output device for taking in text information and taking a photo The entered text information is sent as a picture to the character The character recognition device is configured to identify the text information in the picture and send the text information to the voice conversion device; the voice conversion device is configured to convert the text information into voice data, and send the message information to the a voice output device; the voice output device, configured to play the voice data. The patent application discloses a technique for collecting and recognizing text symbols in image information and then converting the text symbols into speech.

CN102956231A (Publication Date March 6, 2013) discloses a semi-automatic correction based speech key information recording apparatus and method in the field of speech recognition technology, the apparatus comprising: a key information extraction unit and an information correction unit connected thereto, The key information extracting unit obtains the uncorrected text data and extracts the key information, and outputs the key information to the information correcting unit, and the information correcting unit outputs the text data confirmed by the user feedback. The invention reduces the workload of manual correction by using a semi-automatic information correction unit; uses a database to correct special nouns such as place names and professional tool names, thereby reducing the influence caused by the operator's knowledge limit in manual correction; extracting voice data Key information in the message, thereby increasing the amount of information available for the recorded information. The patent application aims to solve the problem of semi-automatic correction of text data after speech conversion into text.

CN105159870A (Publication Date December 16, 2015) discloses a processing system for accurately completing continuous natural speech textualization, the processing system comprising a cloud speech recognition engine and a speech recognition post-correction platform, the speech recognition post-correction platform and The cloud speech recognition engine is connected, and the speech recognition post-correction platform comprises a display unit, a correction operation unit, a control unit and a three-dimensional integrated generation unit, and the correction operation unit comprises a speech correction, a keyboard correction, a mouse correction and a keyboard plus The correct operation mode of the mouse, which discloses that the voice file to be recognized can be finely segmented to achieve accurate recognition.

CN105808197A (Publication Date July 27, 2016) discloses an information processing method applied to an electronic device having a speech recognition module, the method comprising: receiving input speech data; After the voice data is input and the recognition result is obtained, when the first information in the recognition result is the content that needs to be corrected, the first information is at least one character in the recognition result, and the manner of inputting through the operation body is adopted. Correcting the first information in the recognition result, wherein the first information in the recognition result is corrected by the manner of inputting the operation body, and only the part of the target correction is corrected, without the user inputting the voice data again. The objective result can be obtained, the operation process is simple, and the overall speed of information input is improved. The patent application discloses that it is only necessary to correct the content that needs to be corrected at the first position after speech recognition, thereby improving the speed of the correction, but such correction is only for the recognized text data, wherein the process of speech recognition In the middle, the method of comparing the information to be identified with the standard voice data is used, thereby improving the recognition accuracy.

CN106328145A (Publication Date, January 11, 2017) discloses a voice correction method and apparatus, comprising: acquiring voice data input by a user; and identifying the voice data to obtain text content corresponding to the voice data; When the text content includes the first preset keyword, the text content is divided into original text and edit text according to the first preset keyword, wherein the edit text is used to perform the original text. Correction; according to the edited text from the Extracting the text to be corrected in the original text; correcting the original text according to the edited text and the to-be-corrected text to obtain the corrected text. The patent application discloses that the text to be edited in the original text, that is, the edited text, can be obtained by means of keyword recognition, and the correction is made in a targeted manner.

CN102215233A (Publication Date, October 12, 2011) discloses an information system client installed in a user's terminal device, which can be applied to a microblog, a blog, a forum or a personal space, etc., including: a user interaction module and a connection office. The voice module of the user interaction module, preferably, further includes a feedback module, a conversion module, the voice module includes a voice collection unit, a voice recognition unit, and a voice synthesis unit, and the voice collection unit is configured to collect voices of the user; the voice recognition unit The voice recognition unit collects the voice recognition as text output to the user interaction module; the voice synthesis unit converts the text obtained by the user interaction module from the information system server into voice output to the user; the feedback module, the connection center a voice recognition unit, configured to confirm whether the voice recognition is correct, if the correct, the feedback module outputs the text to the user interaction module, and if not, the feedback module enables the voice collection unit Re-acquiring the user's voice or the voice recognition unit corrects the text straight To confirm that it is correct. The patent application discloses a technology for converting between voice and text respectively, and aims to convert information of one format into information of another format, and if the outputted text information is incorrect, the feedback module re- Collect user voices or directly correct the text information of the output.

CN106486113A (March 8, 2017) discloses a method of recording a meeting, comprising: Obtaining a voice signal; converting the voice signal into corresponding text information by a voice conversion software, and displaying the text information, wherein the text information includes correct text information and incorrect text information; and performing error text information in the document Marking, and linking the erroneous text information of the mark to the voice signal corresponding to the erroneous text information; when clicking the erroneous text information, using the voice conversion software to associate the voice linked with the erroneous text information The signal is secondarily recognized, and the second recognized text information is editably displayed in the document; the error text information is corrected and edited in the editable display to obtain corrected text information, and the corrected text information is used Replace the error text message.

In summary, in the prior art, neither the teaching recording and broadcasting field nor the speech recognition conversion field involves the correction concept of the speech itself to be recognized. Everyone cares about the speech recognition conversion, especially the speech conversion into text. The problem of accuracy. However, in any teaching or conference process, for any speaker, there may be misrepresentation, omission or non-standard pronunciation, and even non-standard expression. For these problems, it is usually used in speech recognition. That is, when converting to text (such as in subtitles), plus text annotations (such as explained in parentheses).

In particular, for the teaching and recording system, since the taught course is recorded and reproduced to the user through the network, the influence of the problems such as wrong saying, missing, and non-standard expression becomes prominent and affected because the language data is compressed. Very large, on the one hand, because users often find it difficult to identify these errors, and even if they are identified by subtitles, on the other hand, because of the use of the environment, the user may It is inconvenient to watch subtitles, and can only be listened to in the form of voice. The voice expression is unclear, which further affects the effect of user learning.

In view of the problems existing in the prior art, the present invention aims to provide a teaching recording and reproducing data correcting apparatus, which replaces the text of the specific corrected text with the standard voice data on the basis of correcting the text converted by the voice. The corresponding voice segment corresponding to the corrected text content in the original recorded voice data forms standard voice data and corresponding text, so that when the recorded data is recorded on the spot after the event, the voice data different from the original recorded voice data can be played. Correct voice, and display the correct subtitle information.

Summary of the invention

The invention aims to provide a teaching recording and broadcasting data correction device with a voice correction function, which comprises converting a voice signal in a network teaching or an online conference into original voice data with time stamp using a recording device, and using a voice recognition model. Converting the original voice data into original text data, correcting the original text data, replacing the old text content to be corrected with the new text content, realizing correction of the original text data to form corrected text data, using time stamping Positioning, replacing the standard voice data of the new text content with the corresponding voice data segment of the old text content to form the modified voice data.

It should be understood that although the description mainly describes an embodiment of the present invention in the name of a network teaching recording system or a network conference system, it can be understood that the apparatus of the present invention can also be used for other network online communication processes. Record and play. That is, the invention relates to the provision of online teaching, online training, emergency command (map annotation and voice recording), financial systems or online The method of teaching and recording of the conference system or the method, system and computer program product of the recording and playing process of the conference, in the process of network teaching, online training, emergency command (map annotation and voice recording), financial system (marketing explanation) or online conference As long as the recording of the voice data is involved, the correction of the text data by the conversion of the voice data is recognized, and the standard voice data of the corrected text content is replaced with the corresponding voice data of the original recording, so that the correction of the recorded voice data can be realized. .

The invention provides a teaching recording and broadcasting data correction device, in the process of recording and on-demand review of a multimedia classroom (or network classroom) or the like, especially when recording a multimedia classroom, including voice data and a multimedia whiteboard. The action data (electronic whiteboard book), the operation data on the screen of the user terminal, the video data recorded by the recording device, etc. are added in time stamps in the data stream format and then saved, forming the recorded data. After the user logs in to the network teaching and recording system, the cable is used. Or the wireless local area or the wide area network obtains the recorded data, realizes the reproduction process on the user terminal by using the time stamp or simulates the teaching process of the reproduction classroom, thereby realizing the review playback or the on-demand playback of the recorded classroom.

The teaching recording data correction device of the present invention comprises a file identification generating unit, a voice data collecting unit, a voice data correcting unit, another data collecting unit, a recording data playing unit and an error information feedback unit, wherein

a file identification generating unit, configured to generate a file identification ID when starting the recording teaching process;

a voice data collecting unit, configured to convert a voice signal into original voice data by using an audio collecting device, and save the voice data stream format;

a voice data correction unit, configured to correct voice data that needs to be corrected in the original voice data, to form corrected voice data;

The other data collection unit is configured to collect at least one of the following data: action data on the multimedia whiteboard, operation data on the screen of the user terminal, video data of the video recording device, and adding the timestamp to each data collected, And separately saved in a data stream format, and together with the modified voice data stream and the modified text data, form recordable data that can be played;

Recording a data playing unit, the user uses the terminal to acquire the recorded data through a network, combines different data streams according to the time stamp, thereby playing the recorded data on the terminal, reproduces and/or simulates a recurring teaching process, and realizes Learning and/or reviewing the teaching process;

The error information feedback unit may: when the user plays the recorded data by using the terminal, may select and submit the error text content in the found modified text data, and the feedback content is updated by the administrator, and the correction is updated. Text data, and repeating the voice data replacement unit, updating the modified voice data.

The voice data modification unit further includes a voice data recognition unit, a text data correction unit, and a voice data replacement unit, wherein:

a voice data identification unit, configured to convert the original voice data identification into original text data;

a text data correction unit, configured to correct the original text data, and correct the old text content that needs to be corrected into an accurate new text content to form corrected text data;

And a voice data replacing unit, configured to replace the voice data stream segment of the old text content in the original voice data with standard voice data of the new text content to form a modified voice data stream.

The voice data collecting unit is configured to collect at least one voice data from at least one voice source, add a time stamp, and save the voice data stream format;

The voice data identification unit is configured to convert the voice data stream identification into text data, the text data includes the time stamp, and the time of each text content in the text data may be determined according to the time stamp coordinate.

The voice data replacing unit is configured to retrieve standard voice data of the new text content from a standard voice database, and replace the old one of the original voice data with the standard voice data according to the time stamp A segment of the speech data stream corresponding to the textual content, thereby forming a stream of modified speech data.

The modified text data is displayed on the screen of the terminal in a subtitle manner according to the time stamp, preferably displayed on a screen area in which video data is played, and more preferably, the text data is editable The way is displayed in a specific area of the terminal, in a selectable manner.

When the text data and the voice data are corrected or updated, a correction history record is formed, which may include correction time, correction content, correction operator, problem finder, and the like.

The voice data replacing unit is configured to calculate smoothing according to the pronunciation time of the replaced old text content in the original voice data and the pronunciation time of the standard voice data of the new text content. The coefficient further adjusts the pronunciation time of the new text content according to the smoothing coefficient, thereby causing smoothing and synchronization of the voice data before and after replacement.

The old text content may be empty content, that is, the new text content replacing the empty content is missing, and the text content needs to be added now.

The new text content may be empty content, that is, the old text content that is replaced is redundant, and the deleted text content is now required.

Through the method of the invention, the level of classroom recording is improved, various data are separately saved by means of the identifier of the time stamp, and the voice data is corrected by the recognition and conversion of the voice data and the correction of the text data, and the voice data is corrected according to the corrected text content. The content that needs to be corrected in the original recorded voice data overcomes the problems caused by "less talk, wrong talk and miss talk" in the classroom, and can obtain double corrected speech data and text data (subtitle information).

The above and further objects and features of the present invention will become more apparent from the following detailed description.

DRAWINGS

1 is a block diagram of a recording and broadcasting system according to the present invention;

Figure 2 is a flow chart showing the recording and recording steps in accordance with the present invention;

Figure 3 is a flow chart of speech correction in accordance with the present invention.

Preferred embodiment of the invention

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

The network teaching in the invention is not limited to the classroom teaching form of students and teachers, and may include online network teaching, remote network teaching, local network teaching, and employees of enterprises and institutions, with teachers and students, or trainers as participants. Participate in online web conferencing, remote web conferencing, local web conferencing, and other forms of communication/interaction that use the web for online communication and/or presentation of file content, such as remote collaborative work.

As shown in FIG. 1 , the teacher 1 and the student 2 respectively connect to the teaching server 3 via the Internet using a terminal device installed with a client of the network teaching recording and broadcasting system, thereby realizing network lecture/listening/recording/on-demand/review of the multimedia classroom. .

The terminal device includes: a processor, a network module, a control module, a display module, and a smart operating system, and can be a smart phone, a PAD, a notebook computer, a desktop computer, or the like. The terminal may be provided with a plurality of data interfaces for connecting various extension devices and accessories through a data bus. The intelligent operating system includes Windows, Android and its improvements, iOS, on which application software can be installed and run, and functions of various application software, services, and application stores/platforms under the intelligent operating system are realized.

Terminal devices can be connected to the Internet via RJ45/Wi-Fi/Bluetooth/2G/3G/4G/G.hn/Zigbee/Z-ware/RFID connections and connected to other terminals or other computers and devices via the Internet. Through 1394/USB/Serial/SATA/SCSI/PCI-E/Thunderbolt/data card interface and other data interfaces or bus methods, through HDMI/YpbPr/SPDIF/AV/DVI/VGA/TRS/SCART/Displayport, etc. Audio and video interface, etc. The connection method is used to connect various expansion equipment and accessories to form a conference/teaching equipment interactive system. The sound capture control module and the motion capture control module with software form, or the sound capture control module and the motion capture control module in the form of data bus onboard hardware, realize voice control and shape control function; connect display/projection through audio and video interface Modules, microphones, audio equipment and other audio and video equipment for display, projection, sound access, audio and video playback, and digital or analog audio and video input and output functions; connected to the camera, microphone, electronic whiteboard, RFID through the data interface The reading device realizes image access, sound access, use control and screen recording of the electronic whiteboard, RFID reading function, and can access and control mobile storage devices, digital devices and other devices through corresponding interfaces; through DLNA/ IGRS technology and internet technology are used to implement functions such as manipulation, interaction and screen switching between multi-screen devices.

In the present invention, a processor is defined to include, but is not limited to, an instruction execution system such as a computer/processor based system, an application specific integrated circuit (ASIC), a computing device, or a non-transitory or non-transitory computer. A hardware and/or software system that reads a storage medium to acquire or acquire logic and execute instructions contained in a non-transitory storage medium or a non-transitory computer readable storage medium. The processor may also include any controller, state machine, microprocessor, internetwork-based entity, service or feature, or any other analog, digital, and/or mechanical implementation thereof.

In the present invention, the Internet may include a local area network and a wide area Internet, and may be a wired Internet or a wireless Internet, or any combination of these networks.

As shown in FIG. 2, the main steps of the network teaching recording according to the present invention are as follows:

S100: Starting the recording and broadcasting system: the user uses the terminal to log in, the intelligent electronic whiteboard, the teacher terminal screen operation motion capturing program, the camera, the microphone and other multimedia teaching equipment enter the working state, the camera may have more than one, the microphone includes at least one, respectively Used to capture the teacher's voice and to capture the student's voice, the recording server's teaching server can be used to generate digital timestamps.

S200: Start online teaching: the teacher starts classroom teaching, and the recording and broadcasting system generates a teaching document ID. In the teaching process, for example, the teacher uses the intelligent electronic whiteboard to display (as a teaching board or explain the problem board), and uses real-time voice to explain and use. Real-time interactive voice communication, and can also be displayed and explained on the teacher terminal using electronic documents such as PPT documents, so as to carry out multimedia teaching and interactive question and answer communication with students.

S300: Recording data saving: During the recording process, the action on the intelligent whiteboard is transmitted and saved in the form of “action data stream + time stamp”. The voice in the teaching and interaction process is “voice data stream + time stamp”. Transmission and storage, the operation actions of electronic documents such as PPT documents involved in the teacher terminal are transmitted and saved in the manner of "electronic document operation data stream + time stamp", and the collected video data is transmitted in the form of "video data stream + time stamp". And save. All of these data streams throughout the course of the course are tied to the teaching document ID to achieve the identity of the recorded course. These data can be added or deleted as needed. A typical case is that the recorded data includes voice data, video data, and PPT document presentation data, and the PPT document presentation data can usually be displayed in the form of video data. You must use an action action to reproduce it. In the prior art, classified recording Split screen display is a relatively mature technology. The various data recorded can be saved to a local database or a terminal database, and then uploaded to the remote teaching server through the network, or directly saved to the remote teaching server.

In one example, for the acquisition of voice data, a voice acquisition device, such as various available microphones, can be used to acquire the voice signal, and the voice signal can be converted to voice data for storage in a data stream format. In the case of a single speech source, the gender of the speech source can be marked so that the standard speech of the corresponding gender can be selected for subsequent speech correction (replacement) operations. For the case of multiple voice sources, the gender of the voice source can be separately identified, and the multiple voice sources can be identified, and the time stamps can be separately saved and the multiple voice sources can be separately identified. I will not repeat them here.

S400: Voice data conversion: For the recorded original voice data, the original text data is first formed by the voice model, and then the original text data is corrected. When the original text data is formed, the time stamp of the original voice data is added to the text data so that the text content in the text data can be time-located. The text content may be at least one word, word, sentence or paragraph in the text data. The clock data of the time dimension of the audio data can be obtained by the time positioning, that is, the clock parameter of the time point at which a certain data segment in one audio data can be relatively located.

When performing the recognition conversion, the original speech data identification can be converted into the original text data by using various available speech models, and when the speech data recognition conversion is performed, the gender of the speech source is first recognized. And adding gender information to the text data. Proofreading corrections for text data include manual proofreading, semi-automatic proofreading, and voice proofreading.

S500: Voice data correction: The original text data is corrected using a voice correction command, that is, using a voice proofing method (CN106406807A), but the present invention is not limited thereto. The voice proofreading unit includes: receiving a voice correction instruction, identifying, in the text data to be corrected, all the characters that are the same as the voice correction instruction sound, and a time stamp of the text content, determining the to-be-corrected text in all the recognized texts, and displaying The alternative text list corresponding to the to-be-corrected text accepts an alternative text selection instruction, performs a replacement operation, and forms corrected text data, thereby completing the text correction.

In the process of completing the text correction, the standard pronunciation information of the corrected text is retrieved from the standard speech database, and the corresponding speech data segment is replaced with the standard pronunciation information according to the time stamp of the corrected text to form the corrected speech data. The standard speech database may include a girls standard speech database, a boys standard speech database, and/or a personalized standard speech database. The personalized standard voice database is a voice model of a specific speaker formed by a standard voice database formed by recording a specific speaker, or by corpus training, and can be used for voice recognition, and can also be used to generate personalized standard voice. database.

When the standard pronunciation information is retrieved from the standard voice data, the corresponding standard voice is selected according to the voice source gender information of the original text data, or other personalized information. Alternatively, the old text content may be empty content, that is, the new text content replacing the empty content is missing, and the added text content is now required. The new text content can be empty content, that is, The old text content that was replaced is superfluous and now needs to delete the text content.

As shown in FIG. 3, in one example, the specific steps of the speech correction are as follows:

S11: Receive instruction

When the identified text data finds a problem, if the text to be corrected is “Hu Jian”, the voice correction instruction is received, for example, the user can issue a “selected Hu Jian” voice instruction through the unit to initiate the correction of the problem text “Hu Jian”. instruction.

S12: Find text

All texts identical to the specified pronunciation of the speech correction instruction are identified in the original text data.

S13: Determine the text

Determine all the text to be corrected in the recognized text data.

Wherein, when a plurality of words corresponding to the specified pronunciation of the voice correction instruction appear in the text data, the user can clarify which character needs to be corrected by using a further voice instruction. For example, in the text data to be corrected, the words that are recognized as "hujian" from the time of going to the following are: "Hu Jian", "mutual see", "shoulder shoulder", etc., the user currently wants to recognize the first If a text is corrected, the "first" voice can be issued to determine the first recognized text as the current text to be corrected.

S14: Alternative list

Displaying an alternative text list corresponding to the text to be corrected; the candidate text is homophone with the text to be corrected.

Wherein, when the text to be corrected is selected, an alternative text of the homophone is displayed in the vicinity of the text. A list of words that allows the user to subsequently select alternate text. For example, if the first word "Hu Jian" in the text data is "hujian" is determined as the text to be corrected, then the first word in the text data in this step is "hujian". "A list of alternative texts is displayed nearby: 1, Fujian; 2, accessories; 3, shoulder pads; 4, mutual see,...

S15: Selected instruction

Receive an alternate text selection instruction.

Among them, the user can speak the position of the alternative text in the alternative text list by voice, and complete the work of selecting the alternative text. For example, use Fujian to replace Hu Jian.

S16: Correct text

Correcting the to-be-corrected text to an alternate text specified by the alternate text selection instruction. In the process of performing correction and replacement, the time position information of the text to be corrected is marked with a time stamp, thereby accurately positioning the time position information of the voice data corresponding to the corrected text. Preferably, in the process of correcting the text data and the voice data stream, a correction history record is formed, the correction history record including correction time, correction content, correction operator, and the like.

S17: Voice clip

From the standard speech library, the standard speech data is searched according to the alternative text, and if a plurality of words or sentences are combined, a new piece of speech data is combined. Preferably, the text data includes gender information of the voice source, and when the search is performed, the girl's pronunciation or the boy's pronunciation, or various voice data such as various trebles and basses may be obtained according to the gender information.

S18: Voice replacement

The new voice data segment is replaced with the corresponding voice data segment in the original voice data according to the previously described time position information to form new voice data. Preferably, due to the pronunciation time of the standard speech and the pronunciation time of the replaced speech, even if the text content is completely the same, the pronunciation time is not necessarily the same. For smooth seamless replacement, the pronunciation time of the two speech segments may be calculated first. And a smoothing coefficient, according to the smoothing coefficient, speeding up or slowing down the standard pronunciation time, so that the pronunciation duration of the same text content after the replacement and before the replacement is consistent.

The user uses the terminal to log in to the recording and broadcasting system through the Internet, and can realize the review playback or on-demand playback of the recorded classroom. Of course, for some users, such as online online conference users, these recording classrooms may be process record files of online online conferences, and the recording and playback system will send the teaching file IDs requested by the user for review or on-demand to the teaching server through the Socket encrypted channel, through teaching. The file ID obtains the time-stamped action data stream, the voice data stream, the electronic document operation data stream, the video data stream, and the text data of the course to be sent to the user terminal requesting the corresponding teaching file ID, and the user terminal locally according to the timestamp. Restore (reproduce or simulate reappearance) the entire classroom teaching process. These data streams can be displayed or switched display in each functional area of the user terminal. For video, it can generally be reproduced on the user terminal, but for the operation of the electronic whiteboard, simulation reproduction can be realized by the simulation program of the electronic whiteboard.

Of course, the user can choose to play only at least one of these data streams, for example, can only listen to the voice. For text data, it can be displayed in a specific area of the user terminal in the form of subtitles, such as a video exhibition. In the exhibition area.

In one example, the text data functioning as a caption can be displayed in a specific editable area, so that the user can perform a selected operation or the like, so that only the corresponding text needs to be selected for the found non-standard voice data or text information. Information can be fed back. The administrator of the recording and broadcasting system verifies the feedback after receiving the feedback from the user. If it finds that there is an error, repeats the correction steps of the previous text data and the voice data stream, so that the text data and the voice data can be continuously improved and improved. .

In the above embodiment, the terminal and the server are configured to be connected to a communication network including the Internet. Therefore, the medium may be a program that carries the program code in a streaming manner via the communication network. In the case where the program code is downloaded from the communication network as described above, the program for downloading may be stored in the main device or may be installed from another recording medium. Furthermore, the present invention can be realized by the above-described program code in the form of a computer data signal embodied in an electronic transmission embodied in a carrier wave.

The preferred embodiments of the present invention have been described above, and the present invention is intended to be illustrative and not to limit the scope of the present invention. It is intended to be included within the scope of the appended claims.

Industrial applicability

The teaching and recording data correction device provided by the application improves the level of classroom recording, and saves various data by means of time stamp identification, through recognition and conversion of voice data and correction of text data, and according to the correction Corrected voice data with text content, corrected original recorded language The content that needs to be corrected in the audio data overcomes the problems caused by "less talk, wrong talk and miss talk" in the classroom, and can obtain double corrected speech data and text data (subtitle information).

Claims (10)

A teaching recording and playing data correction device comprises a file identification generating unit, a voice data collecting unit, and a voice data correcting unit, wherein The file identifier generating unit is configured to generate a file identifier ID when starting the recording teaching process; The voice data collecting unit is configured to convert the recorded voice signal into original voice data by using an audio collecting device, and save the voice signal in a voice data stream format; The voice data correction unit is configured to correct voice data that needs to be corrected in the original voice data to form corrected voice data. The apparatus according to claim 1, wherein the voice data correction unit comprises: a voice data identification unit, configured to convert the original voice data identification into original text data; a text data correction unit, configured to correct the original text data, and correct the old text content that needs to be corrected into an accurate new text content to form corrected text data; And a voice data replacing unit, configured to replace the voice data stream segment of the old text content in the original voice data with standard voice data of the new text content to form a modified voice data stream. The device according to claim 2, characterized in that The voice data collecting unit is configured to collect at least one voice data from at least one voice source, add a time stamp, and save the voice data stream format; The voice data identification unit is configured to convert the voice data stream identification into text data, the text data includes the time stamp, and the time of each text content in the text data may be determined according to the time stamp coordinate. The device according to claim 3, characterized in that The voice data replacing unit is configured to retrieve standard voice data of the new text content from a standard voice database, and replace the original with the standard voice data according to the time stamp A segment of the voice data stream corresponding to the old text content in the voice data, thereby forming a modified voice data stream. The apparatus of claim 4 further comprising other data acquisition units, The other data collection unit is configured to collect at least one of the following data: motion data on the multimedia whiteboard, operation data on the screen of the user terminal, video data of the video recording device, and time stamping for each data collected, and Each of them is separately stored in a data stream format, and together with the modified speech data stream and the modified text data, forms recordable data that can be played. The apparatus according to claim 5, further comprising a recording data playing unit, Through the recording data playing unit, the user acquires the recorded data through the network by using the terminal, and combines the acquired different data streams according to the timestamp, thereby playing the recorded data on the terminal, reproducing and/or simulating the weight. The teaching process now enables learning and/or review of the teaching process. The device according to claim 6 wherein The modified text data is displayed on the screen of the terminal in a subtitle manner according to the time stamp, preferably displayed on a screen area in which video data is played, and more preferably, the text data is editable The way is displayed in a specific area of the terminal, in a selectable manner. The apparatus according to claim 7, further comprising an error information feedback unit, The error information feedback unit may: when the user uses the terminal to play the recorded data, select and submit the error text content in the found modified text data, and the feedback content is confirmed by the administrator, and the updated content is updated. The corrected text data is described, and the modified voice data is updated by the voice data replacing unit. The device according to claim 8 wherein When the text data and the voice data are corrected, a correction history record is formed, which may include correction time, correction content, correction operator, problem finder, and the like. The device according to claim 4, characterized in that The voice data replacing unit is configured to calculate a smoothing coefficient according to the pronunciation time of the replaced old text content in the original voice data and the pronunciation time of the standard voice data of the new text content, and according to the smoothing coefficient, The pronunciation time of the new text content is adjusted, thereby smoothing and synchronizing the voice data before and after the replacement.

Images