ITEM

Field of the Invention

[01] The present invention generally relates to a system for recording an interpretation of a source media item by a student. The source media item corresponds to an audio and/or video source item that is streamed from a teacher computer to one or several student computers. The interpretation by the student and the source media item are recorded during playout of the source media item such that the interpretation can be reviewed and evaluated afterwards.

Background of the Invention

[02] In a PC (personal computer) based language translation or interpretation training platform, an audio and/or video source item is streamed from a teacher PC to a plurality of student PCs of respective students that each must interpret the audio and/or video source while it is being played out. The audio and/or video source item that is streamed may be selected by the teacher from a database containing training material or may be selected from the internet. For instance, a teacher may select a Youtube movie that is streamed from the teacher PC to student PCs to be interpreted by the students.

[03] In order to be able to review, discuss and evaluate the interpretations by the different students, the student speech and/or images are recorded during playout of the source item using for instance the microphone and/or webcam of the student PC. Robustness of the recording of the student speech and/or images is of utmost importance in order to be able to reliably and objectively evaluate the student afterwards.

[04] At present, systems are available wherein the source media item that is streamed from the teacher PC to the student PCs is recorded on all student PCs in order to get it synchronised with a local recording of student speech and/or images. Such recordings represent huge files that must be stored temporarily on the student PCs and that must be transmitted back to the teacher PC for evaluation.

[05] Apart from the fact that the recording of the source media item and the recording of the student speech/images represent huge files that consume substantial student PC and network resources, many errors can occur during the recordings and/or during the synchronization of the recordings. Network glitches or artefacts for instance may influence the recording of the source media item at a student PC. Network glitches or artefacts also may hamper the transmission of recordings from a student PC to the teacher PC. Local processes running on the student PCs like for instance firewalls, virus scanners, etc., may restrict the available processing power or storage capacity as a result of which the recording of the source media item or the recording of student speech/images may be delayed or interrupted. The hard disk of a student PC whereon a recording is stored may become occupied for seconds by other processes as a result of which the student speech/images in the recordings may be shifted over seconds, whereas evaluation of an interpretation typically requires lipsync accuracy, i.e. milliseconds accuracy in the synchronization between source media item and interpreter's speech/images. Summarizing, the student PC environment and the network between the teacher PC and student PCs cannot be sufficiently controlled and relied upon for robust source media item and interpretation recording and synchronization.

[06] International patent application WO 2017/106939 A1 from applicant Televic Education NV, entitled "Conference System for the Training of Interpreters", describes a system for training and evaluating student interpreters. The system comprises a teacher unit (125), student units (145a-145c) and a digital network (the so called second digital network in WO 2017/106939 A1 , represented by the dotted lines in Fig. 1 ). The teacher unit and student units can be standard PCs. The teacher unit records the source signal as well as speech/video images captured by and obtained from the student units. The problem of synchronising the recorded source signal with the student speech/images when evaluating the student is mentioned on page 5, lines 26- 32 and page 9, lines 31 -34, but WO 2017/106939 A1 fails to disclose a technical solution to this synchronization problem. Furthermore, WO 2017/106939 A1 assumes on page 12, lines 1 -8, that the network has sufficient capacity. The many network artefacts that can hamper or delay the transmission and/or recording of source signal and student speech/images thus are neglected in WO 2017/106939 A1 .

[07] The Luxemburg patent application LU91549A2 describes another electronic learning system for students of conference interpretation that makes use of personal computers. The system known from LU91549A2 allows a teacher to select a source item from a database and to stream the source item to the student PCs. The source item can be recorded centralized or it can be recorded locally in the student PCs. A student has the option to select synchronized interpretation (speaking while the source media item is played out) or consecutive interpretation (first watching the source media item, then interpreting). In the situation where synchronized interpretation is elected by a student, a speech signal locally recorded by the student PC shall be timestamped in order to enable synchronized playout of the locally recorded speech signal with the source item. The system known from LU91549A2 as mentioned in paragraph [001 1 ] further relies on a dedicated player that mutes one of the audio channels (for instance the PC loudspeaker, the headphone's left ear, the headphone's right ear, etc.) at playback of the source item, and replaces the muted signal with the synchronized recorded student speech for evaluation.

[08] LU91549A2 recognizes the problem of resource consumption at the student PC and mitigates this problem by recording no student video images at the student PCs - only student speech is recorded locally - and by restricting the amount of audio files that is recorded locally at a student PC to 9 (see [0072]). Thereafter, local storage resources must be freed by transferring speech recordings to a central storage.

[09] LU91549A2 however does not seem to recognize possible network or PC problems that may hamper or delay the transmission of source item, and the recording of student speech either locally or centrally. Even if the timestamps allocated in LU91549A2 to local student speech recording enable to synchronize the start of the student speech and the source media item, it is not taught how they enable to continuously synchronize the student speech with the source item as required for objective evaluation of a student. [10] Canadian patent application CA 2 725 177 A1 entitled "Reproducing Apparatus and Method, and Recording Medium" concerns a system and method for reproduction of still images, like for instance a browsable slide show, to which sub audio data is separately added. CA 2 725 177 A1 addresses the problem of preventing interruptions in the reproduction of background music during reverse and forward play of still images that is unavoidable when conventional reproducing technology is used (see pg. 8, lines 22-27). The conventional reproducing technology pointed to by CA 2 725 177 A1 is the mechanism known from MPEG standards to synchronize the reproduction of video data and audio data that are encoded in a single MPEG file. This mechanism is described on pg. 1 -7 of CA 2 725 177 A1 and relies on three timestamps: an ATS that determines at the receiver side when a so called TS packet is forwarded to the decoder for being demultiplexed into a video packet and an audio packet, a DTS that determines when the video packet is decoded, and a PTS that determines when the audio packet is decoded. The PTS is also used to synchronize the reproduction of video data and audio data.

Summary of the Invention

[11] It is an object of the present invention to disclose a system for recording the interpretation of a source media item by one or several student interpreters that mitigates one or several of the above-identified problems of existing solutions. More particularly, it is an objective to disclose a system that allows a more robust recording and synchronization of student speech/images with the source media item that served as task for the interpretation, in order to enable a more reliable and objective evaluation of interpretations of that source media item by students.

[12] According to the present invention, the above object is realized by the system for recording an interpretation of a source media item defined by claim 1 , suitable for training of student interpreters when streaming the source media item from a teacher computer to a plurality of student computers that are network connected to the teacher computer, the system comprising:

- a recording module configured to record student speech and/or images that are captured by a student computer during playout of the source media item in fixed length or fixed size chunks, to generate for each chunk metadata comprising a timestamp and an index of at least one speech sample or image frame recorded in the chunk, and to store the metadata in relation with the chunk; and

- a post-processing module configured to compose the interpretation of the source media item from the chunks, and to add or delete based on the metadata audio samples / video frames while composing the interpretation such that the interpretation is continuously synchronized with the source media item.

[13] Thus, the recording of student speech and/or images, captured by the student computer during playout of the source media item in a so called synchronized interpretation, is done in multiple chunks. The speech may be captured using the built- in microphone of the student PC. In addition to the student speech, images may be captured using for instance the built-in webcam in the student PC. In case of an interpretation in sign language for deaf or hard hearing people, only images may be captured of the student without speech being captured. Each chunk has a predetermined limited length in time, for instance a predetermined amount of seconds, or a predetermined limited size, for instance a predetermined amount of samples. By recording the student speech and/or images in multiple chunks, the risk of losing a complete recording of an interpretation due to a temporary PC failure is eliminated. Each chunk shall be stored as a separate file. During the recording, not only the student speech and/or images are stored in chunks, but also additional metadata is stored in relation to each chunk. The additional metadata at least comprises a timestamp and an index of a speech sample (or of a number of speech samples) or an index of a video frame (or of a number of video frames) recorded in the related chunk. The additional metadata for each chunk further also may comprise a header that contains the recording name, type, status and audio or video specific information like the encoding rate (in bits per second or bps), the sampling rate (in frames per second or fps), the width and height, etc. The recording of student speech and/or images in chunks and the generation and storage of metadata in relation to each chunk are jointly considered to constitute the pre-processing. The timestamp and index will later be used in a post-processing step wherein the recorded audio and/or video chunks comprising the student speech and/or images are merged respectively into a single audio file and/or a single video file that is continuously synchronized with the source media item that served as task for the interpretation. The continuous synchronization is obtained by analysing the timestamps and indexes stored as metadata in relation to the chunks, and by adding or deleting audio samples and/or video frames in between the chunks each time there is a mismatch between the recorded timestamp and index. If for instance 3 speech samples are not captured due to a process occupying the student PC resources, the timestamp stored in relation to the next recorded chunk may be TS=5 whereas the corresponding audio sample index may be l=2. Upon merging the recorded chunks into a single audio file, the timestamp value TS=5 indicates that 3 audio samples must be inserted before the recorded speech sample with index l=2 is added to the audio file. Hence, the post-processing adds or deletes audio samples and/or video frames in order to obtain a recording of the student interpretation with correct length, continuously synchronized with the source media item, and continuously synchronized with other students' interpretations if multiple students simultaneously interpret the streamed source media item.

[14] It is noticed that in the context of the current invention, a speech sample or student speech sample is a value or a set of values representing the student's speech as captured at a point in time by a microphone, for instance the built-in microphone in the student PC. An audio sample is a value or set of values representing sound at a point in time. The audio sample may correspond to student speech but must not necessarily correspond to student speech. It may represent any other sound recorded or artificially composed, like for instance silence.

[15] Similarly, it is noticed that in the context of the current invention, an image, an image frame or student image is a set of digital values representing a picture of the student as captured at a point in time by a camera, for instance the built-in webcam in the student PC. A video frame on the other hand is a set of digital values representing a picture at a point in time. The video frame may correspond to a student image but must not necessarily correspond to a student image. It may represent any other picture recorded or artificially composed, like for instance a blanc picture.

[16] The term continuously synchronized in the context of the current invention implies that the student interpretation composed from the recorded student speech and/or image chunks has an overall length in time that corresponds to the length in time of the source media item, and each recorded fixed length chunk or fixed size chunk in the composed student interpretation is time aligned with the corresponding portion of the source media item, being the portion of the source media item that was played out by the student PC at the point in time the chunk was recorded.

[17] In embodiments of the system for recording an interpretation of a source media item defined by claim 2, the recording module and post-processing module are hosted by the student computer, and the system further comprises:

- an upload module in the student computer configured to upload the interpretation to said teacher computer.

[18] Indeed, the recording module that generates chunks of student speech and/or images, that generates metadata, and that stores the metadata in relation with the chunks may be hosted by the student computer. Alternatively or additionally however, the student speech and/or images captured by the microphone and/or webcam integrated in the student computer may be streamed or uploaded to the teacher computer or another central system that hosts the recording module for the pre- processing or parts of the recording module. Pre-processing the captured student speech and/or images locally in the student computer brings the advantage that the generated chunks are not impacted by any uplink network glitches, but requires that the student computer has sufficient resources available for the pre-processing. Pre- processing the captured student speech and/or images centrally in the teacher computer or in another central system may be less vulnerable to resource shortage when for instance the central system is a performant dedicated system that is configured to have as few resource hungry processes as possible, but may suffer from uplink network glitches during the uploading of student speech and/or images resulting in missing chunks.

[19] Similarly, the post-processing module that composes the student interpretation from the stored chunks and related metadata may be hosted by the student computer. In such case, the student computer must host an upload module that is able to upload the locally generated student interpretation to the teacher computer or another central system. Alternatively however, the post-processing may be organized centrally in the teacher computer or another central system. In the latter situation, the post-processing module in the teacher computer or central system must obtain the chunks and the related metadata either from the centralized pre-processing module, from the decentralized pre-processing modules in the student computers, or from a storage facility like for instance a server or cloud store wherein the chunks and metadata are temporarily stored.

[20] In embodiments of the system for recording an interpretation of a source media item defined by claim 3, each chunk has a length of N video frames, N being a positive integer number above zero.

[21] Thus, preferably, in case student images are recorded while the source media item is played out, the captured student images are recorded in chunks that each contain the same integer amount of N video frames. For instance, each chunk may contain one video frame, that is a single image made by the webcam at a point in time, or - more realistically - each chunk may contain thousand video frames that correspond to thousand consecutive images made by the webcam at thousand consecutive points in time determined by the frame rate of the webcam. A video chunk of one minute for example typically comprises about 900 video frames. In such embodiments, each chunk has the same length in time but not necessarily the same size because the size of video frames depends on the encoding thereof.

[22] In embodiments of the system for recording an interpretation of a source media item specified by claim 4, N is configurable.

[23] Thus, advanced embodiments of the system according to the present invention enable the user or administrator to configure the length of the chunks when expressed as an amount of video frames. The length of the chunks may be selected in trade-off between pre-processing requirements and post-processing requirements. A shorter chunk length increases the pre-processing requirements as more, shorter chunks are created and stored, and more extensive metadata is generated. A corrupted record however will have a shorter length, consequently reducing the post-processing requirements for restauration thereof. Similarly, a greater chunk length decreases the pre-processing requirements as less chunks are created and stored, and less metadata is generated that must be processed afterwards. In case of a corrupted recording however, the post-processing required for restauration of such longer chunk will be higher.

[24] In embodiments of the system for recording an interpretation of a source media item specified by claim 5, each chunk has a length of M audio samples, M being a positive integer number above zero.

[25] Thus, preferably, in case student speech is recorded while the source media item is played out, the captured student speech is recorded in chunks that each contain the same integer amount of M audio samples. For instance, each chunk may contain one million audio samples that correspond to one million consecutive discrete time samples of the sound wave captured by the built-in microphone at one million consecutive points in time determined by the audio sampling rate of the analog-to- digital converter. An audio chunk of 1 minute for example typically contains about 1 .3 million samples. In such embodiments, each chunk has the same length in time and also the same size determined by the sample width or bit depth of the analog-to-digital converter, i.e. the amount of bits per audio sample.

[26] In embodiments of the system for recording an interpretation of a source media item specified by claim 6, M is configurable.

[27] Indeed, advanced embodiments of the system according to the present invention enable the user or administrator to configure the length of the chunks when expressed as an amount of audio samples. The length of the chunks may be selected in trade-off between pre-processing requirements and post-processing requirements. A shorter chunk length increases the pre-processing requirements as more, shorter chunks are created and stored, and more extensive metadata is generated. A corrupted record however will have a shorter length, consequently reducing the post- processing requirements for restauration thereof. Similarly, a greater chunk length decreases the pre-processing requirements as less chunks are created and stored, and less metadata is generated that must be processed afterwards. In case of a corrupted recording however, the post-processing required for restauration of such longer chunk will be higher. [28] In embodiments of the system for recording an interpretation of a source media item defined by claim 7, the video frames added while composing the interpretation are repeated video frames of a previous video frame.

[29] In case the chunks comprise student images, missing chunks or corrupted chunks can be restored during the post-processing by adding video frames such that the overall interpretation remains continuously synchronized with the source media item. In preferred embodiments of the system, a video frame that is added during post- processing corresponds to the previous video frame. This way, the impact on the user experience during review and evaluation of the student interpretation is minimized as the only effect such repeated video frame will have is a standstill of the video recording that forms part of the student interpretation. Moreover, when the amount of repeated frames is limited, the standstill will be unnoticeable for the user.

[30] In embodiments of the system for recording an interpretation of a source media item specified by claim 8, audio samples added while composing the interpretation are silent audio samples.

[31] In case the chunks comprise student speech samples, missing chunks or corrupted chunks can be restored during the post-processing by adding audio samples such that the overall interpretation remains continuously synchronized with the source media item. In preferred embodiments of the system, an audio sample that is added during the post-processing corresponds to a silent sample. This way, the impact on the user experience during review and evaluation of the student interpretation is minimized as the only effect such silent audio sample will have is temporary absence of any sound.

[32] Embodiments of the system for recording an interpretation of a source media item defined by claim 9, further comprise:

- a source recording module in the teacher computer, the source recording module being configured to record the source media item.

[33] Thanks to the pre-processing wherein chunks are generated and metadata comprising a timestamp and index is generated and stored with each chunk, it is no longer necessary to record the source media item in the student computers. The source media item can be recorded centrally, in the teacher computer, and such central recording can at any later point in time be used to evaluate a student interpretation as the student interpretation generated from the chunks and metadata through post- processing is continuously synchronized with such centrally recorded source media item. Hence, the processing and storage capacity requirements in student computers are reduced significantly. Moreover, a more qualitative central recording of the source media item, that is not corrupted through network glitches, can be used during evaluation of a student interpretation. A further advantage is that no locally recorded source media items have to be uploaded from the student computers to the teacher computer before evaluation. Consequently, also the upload bandwidth requirements are reduced.

[34] Embodiments of the system for recording an interpretation of a source media item defined by claim 10, further comprise:

- a synchronizing module in each student computer configured to synchronize the timestamp with a teacher timestamp available at the teacher computer.

[35] Thus, in implementations of the invention wherein the source media item is recorded centrally whereas the student interpretations are recorded locally in the student computers, a network wide time synchronization mechanism ensures that all stored timestamps are synchronized. Each student computer hence must be equipped with a synchronizing module synchronizing its timestamp with the clock or timestamp available at the teacher computer.

[36] In embodiments of the system for recording an interpretation of a source media item specified by claim 1 1 , the synchronizing module is configured to:

- send a first message from the student computer to the teacher computer, the first message comprising a student timestamp Ts derived from a student computer clock;

- receive a second message from the teacher computer, sent in reply to the first message, the second message comprising a teacher timestamp Tt derived from a teacher computer clock; - measure the round-trip time RTT between the sending of the first message and receipt of the second message; and

- determine a synchronization compensation between the student computer clock and the teacher computer clock as Tt - Ts - RTT/2.

[37] Thus, the synchronization may be realised through messaging and timing. The synchronization module in the student computer sends a first message wherein it embeds its timestamp Ts and measures the time RTT between the sending of this first message and the receipt of a reply message from the teacher computer. The measured time RTT corresponds to the round-trip time for a message to travel back and forth between the student computer and teacher computer. If the received message contains the timestamp Tt of the teacher computer, this timestamp was generated and embedded in the reply message at a time Ts + RTT/2 according to the student computer clock. Herein, the assumption is made that the travel time of the first message from the student computer to the teacher computer is equal to the travel time from the teacher computer to the student computer. If the timestamp Tt of the teacher computer extracted from the received message differs from Ts + RTT/2, this difference must be compensated by the synchronization module in the teacher computer. The synchronization module in the student computer in other words applies a synchronization compensation equal to Tt - (Ts + RTT/2) = Tt - Ts - RTT/2. This synchronization compensation is added to its own timestamp Ts and the so corrected timestamp is used to generate the metadata stored in relation to the chunks.

[38] In alternative embodiments of the system for recording an interpretation of a source media item specified by claim 12, the synchronizing module is configured to:

- send a first message from the student computer to the teacher computer, the first message comprising a student timestamp Ts1 derived from a student computer clock at the time of transmission of the first message;

- receive a second message from the teacher computer, sent in reply to the first message, the second message comprising a teacher timestamp Tt2 derived from a teacher computer clock at the time of transmission of the second message and a first difference value deltal determined by the teacher computer as the difference Tt1 - Ts1 between a teacher timestamp Tt1 derived from the teacher computer clock at the time of receipt of the first message by the teacher computer and the student timestamp Ts1 in the first message;

- determine a second difference value delta2 as the difference value Tt2 - Ts2 between the teacher timestamp Tt2 in the second message and a student timestamp Ts2 derived from the student computer clock at the time of receipt of the second message; and

- determine a synchronization compensation between the student computer clock and the teacher computer clock as (deltal + delta2) / 2.

[39] Indeed, the above described synchronisation mechanism based on the round- trip time RTT assumes instant reply of the teacher computer on receipt of the first message from the student computer. In this alternative implementation that does not require instant reply, the synchronization module in the student computer sends a first message wherein it embeds its timestamp Ts. At reception of this first message, the synchronization module in the teacher computer stores the difference between its timestamp Tt and the received student computer timestamp Ts as deltal = Tt - Ts. This difference deltal comprises both the clock difference between teacher and student computer and the travel time of the first message from student computer to teacher computer. The synchronization module of the teacher computer thereupon sends a second message wherein it embeds a new timestamp Tt, along with the calculated difference deltal . At reception of this second message the synchronization module in the student computer stores the difference between the received timestamp Tt and its new timestamp Ts as delta2 = Tt - Ts. This delta2 again comprises the clock difference and the travel time, but since the travel direction is now opposite, the added travel time will influence the computer clock difference inversely. Assuming that the travel time in both directions is equal, the actual computer clock difference between student computer clock and teacher computer clock can be accurately estimated as (deltal +delta2) / 2. Assume for instance a computer clock difference of 10 seconds, and a message travel time of 1 second between student computer and teacher computer. The student computer synchronization module sends a message with timestamp Ts = 10:00:00. The teacher computer synchronization module receives the message at local timestamp Tt = 10:00:10 + 00:00:01 = 10:00:1 1 , and consequently calculates a first difference value deltal = Tt - Ts = 00:00:1 1 . The teacher computer synchronization module sends a reply message with timestamp Tt = 10:00:30 and deltal = 00:00:1 1 embedded therein. The student computer synchronization module receives the message at local timestamp Ts = 10:00:20 + 00:00:01 = 10:00:21 , and consequently calculates a second difference value delta2 = Tt - Ts = 00:00:09. At last, the student computer synchronization module calculates the synchronization compensation as delta = (deltal + delta2) 1 2 = 00:00:10. This synchronization mechanism is insensitive to any delay between receipt of the first message by the teacher computer and transmission of the reply message by the teacher computer.

[40] In addition to a system for recording an interpretation of a source media item as defined by claim 1 , the present invention also concerns a corresponding method for recording an interpretation of a source media item, suitable for training of student interpreters when streaming the source media item from a teacher computer to a plurality of student computers that are network connected to the teacher computer, the method being defined by claim 13, comprising:

- recording student speech and/or images that are captured by a student computer during playout of the source media item in fixed length or fixed size chunks;

- generating for each chunk metadata comprising a timestamp and an index of at least one speech sample or image frame recorded in the chunk;

- storing the metadata in relation with the chunk; and

- composing the interpretation of the source media item from the chunks, and adding/deleting based on the metadata audio samples / video frames while composing the interpretation such that the interpretation is continuously synchronized with the source media item.

Brief Description of the Drawings

[41] Fig. 1 illustrates a PC environment wherein an embodiment of the system for recording interpretations of a source media item by student interpreters according to the present invention has been deployed;

[42] Fig. 2 is a functional block scheme of an embodiment of the system for recording interpretations of a source media item by student interpreters according to the present invention; [43] Fig. 3 illustrates the pre-processing in an embodiment of the system for recording interpretations of a source media item by student interpreters according to the present invention;

[44] Fig. 4 illustrates the post-processing in an embodiment of the system for recording interpretations of a source media item by student interpreters according to the present invention;

[45] Fig. 5 illustrates the pre-processing and post-processing in an embodiment of the system for recording interpretations of a source media item by student interpreters according to the present invention; and

[46] Fig. 6 illustrates a suitable computing system 600 for realizing methods and devices according to embodiments of the invention.

Detailed Description of Embodiment(s)

[47] In Fig. 1 a teacher computer 101 is connected to student computers 1 1 1 , 1 12 and 1 13 via respective links 121 , 122 and 123. The connectivity between teacher computer 101 and student computers 1 1 1 , 1 12, 1 13 may be realised through wireline or wireless Local Area Network (LAN) connections. This may be the case when the teacher and student computers are located in a single room like for example a classroom in a training centre. Alternatively, one or several of the links 121 , 122, 123 realizing the connectivity between teacher computer 101 and respective student computers 1 1 1 , 1 12, 1 13 may partially run over public, wide area networks like for instance the Internet. This may for instance be the case if one or several of the student computers are located at home to enable a student interpreter to participate remotely in a training exercise. During a training exercise, the teacher selects a source media item S, for instance a Youtube movie. This source media item S is streamed to the different student computers 1 1 1 , 1 12, 1 13, as is indicated by the arrows 131 , 132 and 133. As a result of network glitches or artefacts the streamed source media item S arrives in affected form at the respective student computers 1 1 1 , 1 12, 1 13. The first student computer 1 1 1 receives a first version S1 or 141 of the streamed source media item S, the second student computer 1 12 receives a second version S2 or 142 of the streamed source media item S, and the third student computer receives a third version S3 or 143 of the streamed source media item S. The student interpreters using the respective computers 1 1 1 , 1 12 and 1 13 are requested to interpret the source media item while it is played out. The interpretations 11 , I2 and I3 by the student interpreters are recorded and uploaded to the teacher computer 101 as is indicated by the arrows 151 , 152 and 153. As a result of local processes, for example virus scanners, firewalls, etc., running on the student computers 1 1 1 , 1 12, and 1 13, playout of the source media item S1 , S2 and S3 as received by these computers, and recording of the interpretations 11 , I2 and I3 using integrated or connected hardware like a webcam, microphone, headphone, etc., may be interrupted or delayed. Further, the upload of the interpretations 11 , I2 and I3 over the network links 121 , 122 and 122 may again be affected by network glitches or artefacts. For evaluation purposes, the teacher must be able to playout the recorded interpretations 11 , 12, 13 in synchronism with the source media item, either the original version S or the versions S1 , S2, S3 received by the respective student computers 1 1 1 , 1 12, 1 13 and used there as basis for the respective interpretations 11 , 12 and I3. Objective evaluation of the interpretations 11 , 12, 13 requires synchronisation accuracy in the range of milliseconds. As a result of the many possible sources of artefacts mentioned here above, the student PC environment illustrated by Fig. 1 cannot be relied upon for robust source item streaming, and interpretation recording and synchronization. For this reason, a system for recording interpretations of a source media item according to the invention has been deployed in the PC network environment of Fig. 1 . The functional components of this system are illustrated by Fig. 2.

[48] Fig. 2 shows a pre-processor 201 or recording module, and a post-processor 202 that jointly constitute an embodiment of the system for recording an interpretation of a source media item according to the present invention. Fig. 2 further shows a camera 203 and microphone 204, the outputs of which are used by the pre-processor 201 . The camera 203 and microphone 204 may for instance correspond to the built-in camera of student computer 1 1 1 and the built-in microphone of student computer 1 1 1 . Fig. 2 further also shows a chunk storage 205. This chunk storage 205 represents a memory wherein information produced by the pre-processor 201 is stored. The chunk storage 205 may for instance correspond to the hard disk that forms part of student computer 1 1 1 , the hard disk that forms part of the teacher computer 101 , or any other memory integrated in or connected to the student computer 1 1 1 or any other memory integrated in or connected to the teacher computer, like for instance an USB disk, a server, a cloud storage, etc. It is noticed that the chunk storage 205 is used by the system according to the present invention but not necessarily forms part of the system.

[49] The pre-processor 201 , also named recording module throughout the text and claims, comprises a chunk generator 21 1 , a metadata generator 212, a timestamp unit 213, an index unit 214 and a synchronizing unit 215. The chunk generator 21 1 is configured to receive the output streams from the camera 203 and microphone 204, and to divide these streams in chunks which are segments of fixed length or fixed size for storage in the chunk storage 205. The metadata generator 212 interacts with the chunk generator 21 1 and is configured to generate metadata for each chunk that is also stored in the chunk storage 205. The metadata composed by the metadata generator 212 for each chunk at least comprises a timestamp received from the timestamp unit 213 and an index received from the index unit 214. The timestamp generated by the timestamp unit 213 is network synchronized, meaning it is synchronized between the teacher computer and student computers, through synchronization unit 215. The index generated by the index unit 214 may for instance be an index number of the first audio sample received from microphone 204 and integrated in a chunk, or it may be an index number of a video frame received from camera 203 and integrated in a chunk.

[50] The post-processor 202 comprises a metadata analyser 221 and an addition/deletion unit 222. The metadata analyser 221 is configured to obtain chunks and their associated metadata from the chunk storage 205. The addition/deletion unit 222 uses the output of the metadata analyser 221 to compose the student interpretation, and thereto adds and/or deletes audio samples or video frames to the chunks extracted from the chunk storage in order to produce an interpretation that is continuously synchronized with the source media item that served as basis for the interpretation. [51] It is noticed that both the pre-processor 201 and the post-processor 202 may be located in the teacher computer 101 . Preferably however, at least the pre-processor 201 is hosted by the student computers 1 1 1 , 1 12, 1 13. The post-processor 202 then may be hosted by the teacher computer 101 , or it may be hosted by the student computers 1 1 1 , 1 12, 1 13 depending on the available processing resources in these respective computers.

[52] Fig. 3 illustrates in more detail the operation of the pre-processor 201 . It is assumed that the pre-processor 201 is hosted by student computer 1 1 1 where source media item S1 is received and played out. While source media item S1 is played out, the pre-processor 201 is recording the source media item S1 as well as student images CAM captured by the webcam integrated in student computer 1 1 1 and student speech MIC captured by the built-in microphone of student computer 1 1 1 . The source media item S1 is split into chunks, S1 C1 or 31 1 , S1 C2 or 312, S1 C3 or 313, S1 C4 or 314, ..., S1 Cn or 31 n. These chunks each have the same length in time, for instance 1 minute playout time of the source media item S1 at student computer 1 1 1 . For each of the chunks 31 1 ... 31 n, the pre-processor 201 generates associated metadata. The metadata consists of a timestamp locally generated in student computer 1 1 1 at the time the associated chunk is recorded and an index indicative for the video frames that are recorded in the associated chunk. This way, metadata S1 M1 or 321 is generated in association with the first source media item chunk 31 1 , metadata S1 M2 or 322 is generated in association with the second source media item chunk 312, metadata S1 M3 or 323 is generated in association with the third source media item chunk 313, metadata S1 M4 or 324 is generated in association with the fourth source media item chunk 314, ..., and metadata S1 Mn or 32n is generated in association with the n'th source media item chunk 31 n. The source media item chunks 31 1 ... 31 n are stored locally in the student computer 1 1 1 in n different files. Also the generated metadata 321 ... 32n are stored locally in the student computer 1 1 1 in n different files. It is noticed that in an alternative implementation, a chunk and its associated metadata may be stored in a single file instead of in two separate files. In a similar fashion, the student images CAM are split into chunks, C1 C1 or 331 , C1 C2 or 332, C1 C3 or 333, C1 C4 or 334, ..., C1 Cn or 31 k by the pre-processor 201 . These chunks also have the same length in time, i.e. 1 minute in the above example. For each of the chunks 331 ... 33k, the pre-processor 201 generates associated metadata. The metadata consists of a timestamp locally generated in student computer 1 1 1 at the time the associated chunk is recorded and an index indicative for the video frames that are recorded in the associated chunk. This way, metadata C1 M1 or 341 is generated in association with the first student image chunk 331 , metadata C1 M2 or 342 is generated in association with the second student image chunk 332, metadata C1 M3 or 343 is generated in association with the third student image chunk 333, metadata C1 M4 or 344 is generated in association with the fourth student image chunk 334, ..., and metadata C1 Mk or 33k is generated in association with the k'th student image chunk 33k. It is noticed that the amount k of student image chunks may differ from the amount n of source media item chunks because the student computer 1 1 1 may be prevented from recording certain chunks due to various processes running on student computer 1 1 1 and occupying the resources thereof. The student image chunks 331 ... 33k are stored locally in the student computer 1 1 1 in k different files. Also the generated metadata 341 ... 34k are stored locally in the student computer 1 1 1 in k different files. Again, it is noticed that in an alternative implementation, a chunk and its associated metadata may be stored in a single file instead of in two separate files. Further similarly, the student speech MIC is split into chunks, M1 C1 or 351 , M1 C2 or 352, M1 C3 or 353, M1 C4 or 354, ..., M1 Ci or 35i by the pre-processor 201 . These chunks also have the same length in time, i.e. 1 minute. For each of the chunks 351 ... 35i, the pre-processor 201 generates associated metadata. The metadata consists of a timestamp locally generated in student computer 1 1 1 at the time the associated chunk is recorded and an index indicative for the audio samples that are recorded in the associated chunk. This way, metadata M1 M1 or 361 is generated in association with the first student speech chunk 351 , metadata M1 M2 or 362 is generated in association with the second student speech chunk 352, metadata M1 M3 or 363 is generated in association with the third student speech chunk 353, metadata M1 M4 or 364 is generated in association with the fourth student speech chunk 354, ..., and metadata M1 Mi or 36i is generated in association with the i'th student speech chunk 35i. It is noticed that the amount i of student speech chunks may differ from the amount k of student image chunks and/or from the amount n of source media item chunks because the student computer 1 1 1 may be prevented from recording certain chunks due to various processes running on student computer 1 1 1 and occupying the resources thereof. The student speech chunks 351 ... 35i are stored locally in the student computer 1 1 1 in i different files. Also the generated metadata 361 ... 36i are stored locally in the student computer 1 1 1 in i different files. Again, it is noticed that in an alternative implementation, a chunk and its associated metadata may be stored in a single file instead of in two separate files.

[53] Fig. 4 illustrates the operation of the post-processor 202. It is assumed that the post-processor 202 is hosted by student computer 1 1 1 and is operating on the source media item chunks 31 1 ... 31 n and their respective associated metadata files 321 ... 32n, the student image chunks 331 ... 33k and their respective associated metadata files 341 ... 34k, and the student speech chunks 351 ... 35i and their respective associated metadata files 361 ... 36i. The metadata analyser 221 in the post-processor 202 analyses the metadata files 321 ... 32n. Upon analysis of the metadata files 321 ... 32n, the addition/deletion unit 222 adds or deletes samples or frames and combines the samples or frames from the source media item chunks 31 1 ... 31 n with the added/deleted samples or frames into a single file 41 1 . In the example of Fig. 4, it is assumed that samples or frames are added to or deleted from the n chunks S1 C1 ... S1 Cn recorded by the pre-processor 201 as a result of which the source media item file 41 1 is composed. Upon analysis of the metadata files 341 ... 34k, the addition/deletion unit 222 adds or deletes frames to or from the student image chunks 331 ... 33k as a result of which another single file 431 is composed. The so composed file 431 has a length equal to the length of file 41 1 , The added frames represent repetitions of the respective frames after which they have been added, and the frames have been inserted at positions in the file 431 such that continuous synchronization is realized between file 431 and file 41 1 . In other words, frames at corresponding positions in the files 41 1 and 431 were recorded with identical timestamps by the pre- processor 201 such that when played out synchronously, the student images recorded while a certain portion of the source media item was played out, are visualized simultaneously enabling an objective evaluation of the student. Upon analysis of the metadata files 361 ... 36i, the addition/deletion unit 222 adds or deletes samples to or from the student speech chunks 351 ... 35i as a result of which yet another single file 451 is composed. The so composed file 451 has a length equal to the length of files 41 1 and 431 . The added samples represent silent samples, i.e. samples filled with silent audio samples, and the added samples have been inserted at positions in the file 451 such that continuous synchronization is realized between file 451 and files 41 1 and 431 . In other words, samples or frames at corresponding positions in the files 41 1 , 431 and 451 were recorded with identical timestamps by the pre-processor 201 such that when played out synchronously, the student images and student speech recorded while a certain portion of the source media item was played out, are visualized and made audible simultaneously enabling an objective evaluation of the student. The locally generated source media item file 41 1 , student image file 431 and student speech file 451 combinedly form the student interpretation 11 that is uploaded in Fig. 1 from the student computer 1 1 1 to the teacher computer 101 . In a similar way, interpretation I2 is generated locally in student computer 1 12 and uploaded to teacher computer 101 and interpretation I3 is generated locally in student computer 1 13 and uploaded to teacher computer 101 . All interpretations 11 , I2 and I3 are continuously synchronized thanks to the network wide synchronized timestamp that is used by the respective instantiations of the pre-processor 201 in the student computers 1 1 1 , 1 12, 1 13.

[54] Fig. 5 illustrates the operation of the pre-processor 201 and post-processor 202 in a specific example situation wherein it is assumed that the pre-processor 201 and post-processor 202 are hosted by the student computer 1 1 1 , a source media item S1 is received by that student computer 1 1 1 and recorded together with student speech MIC captured by the built-in microphone of student computer 1 1 1 while the source media item S1 is played out. The pre-processor 201 splits the source media item S1 into chunks S1 C1 or 51 1 , S1 C2 or 512, S1 C3 or 513, ..., and S1 Cn or 51 n. These chunks 51 1 -51 n contain portions of the source media item S1 of equal length in time, e.g. the length in time of playout of a single video frame by the student computer. Note that this is a simplified, theoretic example as in real implementations it may be expected that the length of a chunk shall span multiple video frames, for instance have a length of 1 minute corresponding to approximately 1000 video frames. For each of the chunks 51 1 -51 n, metadata is generated consisting of a timestamp TS generated by the timestamp unit 213 at the time the corresponding chunk is played out, and an index I generated by index unit 214 and representing a sequence number of the video frame in the corresponding chunk. In Fig. 5 it is assumed that the received source media item S1 is divided in chunks 51 1 -51 n and that these chunks are played out by the student computer 1 1 1 without interruptions. Consequently, the metadata 521 generated in association with the first chunk 51 1 comprises a timestamp TS=1 and an index 1=1 , the metadata 522 generated in association with the second chunk 512 comprises a timestamp TS=2 and an index l=2, the metadata 523 generated in association with the third chunk 513 comprises a timestamp TS=3 and an index l=3, ... and the metadata 52n generated in association with the n'th chunk 51 n comprises a timestamp TS=n and an index l=n. In a similar way, the pre-processor 201 splits the captured student speech MIC into chunks M1 C1 or 551 , M1 C2 or 552, M1 C3 or 553, ..., and M1 Ci or 55i. These chunks 551 -55i also have the same length in time, for instance a predetermined amount of audio samples corresponding to the length in time of a video frame. For each of the chunks 551 ... 55i, the pre-processor 201 generates respective metadata 561 ... 56i. The metadata 561 ... 56i consists of a timestamp value TS generated by timestamp unit 213 at the time the corresponding chunk was captured, and an index number I generated by index unit 214 and corresponding to a sequence number of the first audio sample recorded in the corresponding chunk. In Fig. 5 it has been assumed that certain audio samples have not been recorded, for instance because the student computer 1 1 1 was running short in resources. The first chunk 551 was recorded at time TS=1 and its first audio sample has the index 1=1 . At time TS=2 no audio samples are recorded. The second chunk 552 consequently was recorded at time TS=3 and its first audio sample has the index l=g. At times TS=4 and TS=5 no audio samples were recorded. The third chunk 553 consequently was recorded at time TS=6 and its first audio sample has the index l=h. The i'th chunk M1 Ci was recorded at time TS=n, i.e. the time where also the last chunk 51 n of the source media item was played out, and its first audio sample has the index i. Although not illustrated by Fig. 5, the chunks 51 1 ... 51 n of the source media item S1 and their respective associated metadata 521 ... 52n are stored in chunk storage 205 as separate files. Similarly, the chunks 551 ... 55i of the student speech MIC and their associated metadata 561 ... 56i are stored in chunk storage 205 as separate files. The post processor 202 extracts the chunks and their associated metadata from the chunk storage 205. The metadata analyser 221 in the post-processor 202 analyses the metadata 521 -52n and concludes therefrom that no video frames are missing in the recorded chunks for the source media item. The metadata analyser 221 consequently instructs the addition/deletion unit 222 to compose a single file 51 1 from the recorded chunks 51 1 -51 n without addition/deletion of any video frames. The metadata analyser 221 in the post-processor 202 also analyses the metadata 561 -56i and concludes therefrom that audio samples are missing at times TS=2, TS=4, TS=5, etc. The metadata analyser 221 consequently instructs the addition/deletion unit 222 to compose a single file 551 from the recorded student speech chunks 551 -55n wherein silent audio samples CO are inserted at time TS=2, i.e. between the samples M1 C1 and M1 C2, silent samples CO are inserted at time TS=4 and TS=5, i.e. between the samples M1 C2 and M1 C3, etc. It is further noticed that also inside the chunks silent audio samples may be added where student speech samples are missing and/or corrupted speech samples may be deleted. This way, an audio file 551 is composed that has a length in time equal to the length in time of media file 51 1 , and wherein all recorded student speech samples are continuously synchronized with portions of the media file 51 1 that were played out at the time these student speech samples were recorded. The student computer 1 1 1 uploads the so composed media file 51 1 and audio file 551 to the teacher computer 101 for evaluation of the student interpreter.

[55] Fig. 6 shows a suitable computing system 600 enabling to implement embodiments of the system and method for recording an interpretation of a source media item according to the invention. Computing system 600 may in general be formed as a suitable general-purpose computer and comprise a bus 610, a processor 602, a local memory 604, one or more optional input interfaces 614, one or more optional output interfaces 616, a communication interface 612, a storage element interface 606, and one or more storage elements 608. Bus 610 may comprise one or more conductors that permit communication among the components of the computing system 600. Processor 602 may include any type of conventional processor or microprocessor that interprets and executes programming instructions. Local memory 604 may include a random-access memory (RAM) or another type of dynamic storage device that stores information and instructions for execution by processor 602 and/or a read only memory (ROM) or another type of static storage device that stores static information and instructions for use by processor 602. Input interface 614 may comprise one or more conventional mechanisms that permit an operator or user to input information to the computing device 600, such as a keyboard 620, a mouse 630, a pen, voice recognition and/or biometric mechanisms, a camera, etc. Output interface 616 may comprise one or more conventional mechanisms that output information to the operator or user, such as a display 640, etc. Communication interface 612 may comprise any transceiver-like mechanism such as for example one or more Ethernet interfaces that enables computing system 600 to communicate with other devices and/or systems, for example with other computing devices 681 , 682, 683. The communication interface 612 of computing system 600 may be connected to such another computing system by means of a local area network (LAN) or a wide area network (WAN) such as for example the internet. Storage element interface 606 may comprise a storage interface such as for example a Serial Advanced Technology Attachment (SATA) interface or a Small Computer System Interface (SCSI) for connecting bus 610 to one or more storage elements 608, such as one or more local disks, for example SATA disk drives, and control the reading and writing of data to and/or from these storage elements 608. Although the storage element(s) 608 above is/are described as a local disk, in general any other suitable computer-readable media such as a removable magnetic disk, optical storage media such as a CD or DVD, - ROM disk, solid state drives, flash memory cards, ... could be used. It is noticed that the entire method according to the present invention can be executed centralized, e.g. on a server in a management centre or in a cloud system, or it can be partially executed on a remote electronic device, e.g. worn by the user, and partially on a central server. Computing system 600 could thus correspond to the processing system available centrally or the processing system available in the electronic device.

[56] Although the present invention has been illustrated by reference to specific embodiments, it will be apparent to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied with various changes and modifications without departing from the scope thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of the claims are therefore intended to be embraced therein. In other words, it is contemplated to cover any and all modifications, variations or equivalents that fall within the scope of the basic underlying principles and whose essential attributes are claimed in this patent application. It will furthermore be understood by the reader of this patent application that the words "comprising" or "comprise" do not exclude other elements or steps, that the words "a" or "an" do not exclude a plurality, and that a single element, such as a computer system, a processor, or another integrated unit may fulfil the functions of several means recited in the claims. Any reference signs in the claims shall not be construed as limiting the respective claims concerned. The terms "first", "second", third", "a", "b", "c", and the like, when used in the description or in the claims are introduced to distinguish between similar elements or steps and are not necessarily describing a sequential or chronological order. Similarly, the terms "top", "bottom", "over", "under", and the like are introduced for descriptive purposes and not necessarily to denote relative positions. It is to be understood that the terms so used are interchangeable under appropriate circumstances and embodiments of the invention are capable of operating according to the present invention in other sequences, or in orientations different from the one(s) described or illustrated above.