FR2907935A1 - Synchronization mark detecting method for audio sequence recording system, involves determining data, representing occurring instant of synchronization mark during time window, from data received during time window - Google Patents

Abstract

The method involves receiving successive digital audio data (S-D) representing a digitized audio sequence. Actuation of a switch (26) of an audio sequence recording system (5) by an operator during reception of the data is detected, and a time window is fixed based on the detection. Data representing an occurring instant of a synchronization mark during the time window is determined from the former data received during the time window. An independent claim is also included for a system for detecting a synchronization mark of a digitized audio sequence.

Description

B7973 1 SYSTEM FOR DETECTING SYNCHRONIZED SOUND TRACKS BETWEEN

  FIELD OF THE INVENTION The present invention relates, in general, to sound recording systems. The invention relates more particularly to systems in which the sound is stored on a digital medium. BACKGROUND OF THE INVENTION The present invention relates to sound recording for a film in which audio sequences and picture scenes are recorded separately. For such a film, there is provided a editing step which consists in assembling end-to-end short scenes of images (generally of a duration of between 1 and 3 minutes) filmed individually, and to associate therewith the corresponding audio sequences. . In particular, a perfect synchronization between each scene of images and the corresponding audio sequence must be obtained. It is then necessary to have synchronization marks on the audio sequences and on the scenes of images. To do this, it is generally used, during filming, a hinged board on which is inscribed the number of the plane and taking turns. At the beginning of each take, the wafer is filmed at the same time as an operator announces the setting, then a dry noise is produced by banging the two articulated parts. The announcement and the clap (called "clap") are thus recorded at the beginning of an audio sequence and filmed at the beginning of the corresponding image scene. Since the jacks are short, several audio sequences are recorded one after the other and are usually digitized for ease of processing. Also, a plurality of sockets may be printed on a reel of film that typically contains about 10 minutes of film. A problem that arises then is to detect, in audio sequences, the successive claps corresponding to the different scenes in order to be superimposed during editing, images and sound synchronously. The detection of synchronization marks (claps) on the audio sequences is generally performed after the recording of all the audio sequences. Conventionally, a graphical representation of the audio sequences is used on a screen of a computer system, for example a computer, to locate and mark synchronization claps. This graphical representation generally corresponds to the waveform of the sound track. On the waveform, a clap is marked as a sudden saturation sound that follows a blank before the decision.

  Conventional systems display the waveform of a digitized sound track as a line on which the waveform scrolls. The operator monitors the waveform and stops scrolling when he believes he has detected a sync mark. However, it is necessary to ensure that it is a synchronization mark and not a sound after a period of silence in a socket. To do this, the processing system generally makes it possible to trigger a sound reproduction from a time corresponding to a location chosen by the operator on the waveform. When a marker is definitely detected by the operator, the keyboard of the computer system is used to enter an identifier of the synchronization mark which is then recorded in the computer system associated with the audio sequence concerned.

A disadvantage of such a method of determining timing marks is that it is generally implemented well after the recording of the audio sequences. As a result, for each audio sequence, the operator does not know at which places in the audio sequence he should look for the sync mark. Indeed, although in most cases a clap is performed at the beginning of an audio sequence, in some cases the clap can be performed only at the end of the audio sequence. In addition, in certain cases, for example in the case of shooting a report, it is not possible to make a clap by means of an articulated board. In this case, a particular event occurring during the recording of the audio sequence, for example the slamming of a door, is used as a clapboard. It may then be difficult for the operator to identify the particular sound event that has been used as a clapper. In addition, the same audio sequence may include several claps. Indeed, several scenes of images can be filmed simultaneously with the recording of an audio sequence. For example, if the hinged board used to make the clapboard can not be visible simultaneously by all the cameras, it is then necessary to successively place the board in the field of view of each camera and to realize a clap. When determining synchronization marks, an operator can then make errors in the number of claps to be detected.

In general, the determination of the synchronization marks after the recording of the audio sequences is particularly long and can lead to errors in the detection of synchronization marks.

SUMMARY OF THE INVENTION The present invention is directed to a method of detecting synchronization marks on digital files containing audio sequences that can be implemented during the recording of audio sequences. According to another object of the present invention, the detection of the synchronization marks requires only a few additional actions on the part of the operator performing the recording of the audio sequences.

The present invention also provides a system for recording audio sequences in digital form which makes it possible, during the recording of audio sequences, to detect synchronization marks on the digital files containing the audio sequences.

To achieve all or part of these and other objects, one aspect of the present invention provides a method of detecting at least one synchronization mark of a digitized audio sequence, comprising receiving first successive digital audio data. representative of the digitized audio sequence and comprising the steps of: (a) detecting the operation of a control means by an operator during the reception of the first data; (b) setting a time window in relation to said detection; and (c) determining, from the first data received during said time window, a second datum representative of an instant of occurrence of the synchronization mark during said time window.

According to an exemplary embodiment, the method comprises storing third digital audio data obtained from the first data and forming the digitized audio sequence and storing the second data with the third data.

According to an exemplary embodiment, the detection of the actuation of the control means corresponds to the end of the time window. According to an exemplary embodiment, the method further comprises the following steps: (d) determining, for each successive set of a first number of first data received, a fourth datum representative of the maximum variation of the first data of said set and a fifth datum representative of the instant at which said variation is maximum; and (e) storing, for a second number of the last received sets, the second number of fourth associated data and the second number of fifth data associated, in step (c), the second data item corresponds to the fifth data item; associated with the fourth maximum data of the second number of fourth data. According to an exemplary embodiment, step (d) consists in determining, for each set, the difference between each first datum of said set and the first preceding datum, the fourth datum corresponding to the maximum deviation. Another aspect of the present invention provides a system for detecting at least one synchronization mark of a digitized audio sequence, comprising means for receiving first successive digital audio data representative of the digitized audio sequence, a control means capable of to be actuated by an operator during the reception of the first data, and a means for determining, from the first data received during a fixed time window in relation to the actuation of the control means, a second data item. representative of a moment of occurrence of the synchronization mark during said time window. According to an exemplary embodiment, the system further comprises means for providing third digital audio data obtained from the first data and forming the digitized audio sequence, and a first means for storing the third data and data. the second data. According to an exemplary embodiment, the system further comprises means for determining, for each successive set of a first number of first data received, a fourth datum representative of the maximum variation of the first data of said set and of a fifth datum representative of the instant at which said variation is maximum, and a second memory means, for a second number of the last received sets, the second number of fourth associated data and the second number of fifth associated data, the means for determining the second data determining the fifth data associated with the fourth maximum data among the second number of fourth data. According to an exemplary embodiment, the means for determining the fourth datum is adapted to determine, for each said set, the difference between each first datum of said set and the first preceding datum, the fourth datum corresponding to the maximum deviation. According to an exemplary embodiment, the second storage means is a first-in, first-out memory. BRIEF DESCRIPTION OF THE DRAWINGS These and other objects, features, and advantages of the present invention will be set forth in detail in the following description of a particular non-limiting exemplary embodiment in connection with the accompanying figures among which: Fig. 1 shows, in block diagram form, an exemplary embodiment of an audio sequence recording system for detecting synchronization marks during the recording of the audio sequences; FIG. 2 illustrates first steps of an exemplary method of detecting synchronization marks; FIG. 3 illustrates an exemplary data format used in the implementation of the exemplary method illustrated in FIG. 2; Figure 4 schematically shows an exemplary waveform corresponding to a sound signal; FIG. 5 illustrates another example of data format used during the implementation of the exemplary method illustrated in FIG. 2; and FIG. 6 illustrates second steps of example 10 of synchronization mark detection method illustrated in FIG. 2. Detailed Description For the sake of clarity, the same elements have been designated with the same references in the various figures.

An exemplary method of detecting synchronization marks according to the invention consists in causing the operator who manipulates a recording system of an audio sequence to indicate to the recording system during the recording of the sequence. audio, a time window in which it is safe, or practically safe, that a clapboard has been recorded or will be recorded. The recording system then automatically detects a synchronization mark in the portion of the audio sequence corresponding to the time window delimited by the operator. The operator being present 25 during the production of the clap and indicating to the recording system, by a voluntary action, the time window in which the detection of a synchronization mark must be performed, thus limiting the risks of false detections synchronization marks. In addition, the indication of the time window can be obtained by operation of a switch of the recording system by the operator. The detection of the synchronization mark in the time window being then performed automatically by the recording system, the number of actions to be performed by the operator is limited.

2907935 B7973 8 For the indication of the time window, three possibilities are possible. The first possibility is, during the recording of the audio sequence, to cause the operator to operate a switch to indicate the beginning of the time window, before the occurrence of a clap, and to operate the switch again. to indicate the end of the time window, after the occurrence of the clap. The second possibility is, during the recording of the audio sequence, to cause the operator to operate the switch once to indicate the beginning of the time window, the duration of the time window being predetermined. The third possibility is, during the recording of the audio sequence, to cause the operator to operate the switch once to indicate the end of the time window, the duration of the time window being predetermined. Although each of these three possibilities can be implemented by the method of detecting synchronization marks according to the invention, the third possibility is the most advantageous. In fact, compared with the second and third possibilities, the first possibility is more demanding on the operator since it requires two actuations of the switch. For the second possibility, the operator must operate the switch while the clap has not yet been realized. There is a risk that it will operate the switch too early and therefore the time window on which the recording system must perform the detection of the sync mark does not contain the clap. For the third possibility, the actuation of the switch by the operator being performed shortly after the occurrence of the clap, it is almost certain that the time window in which the recording system performs the detection of the synchronization mark contains the clap . FIG. 1 represents, in block form, an exemplary embodiment of a system for recording 35 audio sequences in digital form adapted to perform the detection of synchronization marks. The recording system 5 comprises a sound sensor 10, for example a micro-phone, which provides an analog signal SA representative of a sound. An analog-to-digital converter 12 (A / D) receives the analog signal SA and provides a digital signal SD corresponding to a sampling of the signal SA at a sampling frequency F. A preprocessing circuit 14 (PPC), corresponding for example to an FPGA (Field Programmable Gate Array) receives the 10 SD digital signal. The preprocessing circuit 14 may receive other digital signals representative of sound signals. By way of example, the preprocessing circuit 14 can receive a digital signal SD 'from an access terminal IN of the recording system 5, to which is connected a system for providing a sound signal in the form of digital. The pretreatment circuit 14 is partly controlled by a microprocessor 16 (pP) and is adapted to provide, from the samples of the SD signal, PACK data packets which are stored in a memory 18 (RAM), for example a RAM. The microprocessor 16 is adapted to sequentially read each PACK data packet stored in the memory 18, perform certain operations on the read PACK data packet, and produce DATA data which is stored in a memory (FIFO), for example, a first-in-first-out memory and SF data stored in another memory (HD), for example a hard disk. In addition, the microprocessor 16 is connected to a display screen 24 (DISPLAY) and receives a control signal C from a switch 26. An example of a method for detecting synchronization marks that can be described will now be described. implemented by the recording system 5. This method comprises first steps which are carried out continuously as soon as an audio sequence is recorded and second steps which are implemented, during the recording, only when An example of a method for detecting synchronization marks is illustrated in FIG. 2 which illustrates the first steps of the exemplary method of detecting synchronization marks. For reasons of simplicity, it is considered that the preprocessing circuit 14 receives only one digital signal SD representative of a sound signal. By way of example, the digital signal SD corresponds to a series of binary data, coded for example on 24 bits, and provided at a frequency F of 48 10 KHz. Although the steps of the present example process will be successively described for the processing of a portion of an audio sequence, it is clear that they are actually performed, at least in part, simultaneously during the processing of the audio. entire audio sequence.

In step 40, the sensor 10 provides the analog signal SA whose amplitude is representative of a sound level. In step 42, as the analog-to-digital converter 12 receives the analog signal SA, it converts it into a digital signal SD.

In step 44, the preprocessing circuit 14 receives the successive samples of the signal SD and determines, when it has received a number N of samples, for example 2048 samples, a packet of data PACKi, where i is a number positive integer. For example, i is 1 for the first data packet provided by the preprocessing circuit 14, equal to 2 for the second data packet provided by the preprocessing circuit 14, and so on. Figure 3 schematically shows an example of the format of each PACKi data packet. The data packet PACKi comprises: a packet identifier Idi which is, for example, equal to the index i; a group of Sound-Datai data corresponding, for example, to N successive samples of the signal SD; and a group of data DATAi corresponding to two binary data OMaxi and Posi. The OMaxi datum corresponds to the value of the maximum difference between two successive samples among the N 5 samples associated with the PACKi data packet. The data Posi is representative of the position of the sample in the packet PACKi for which the difference between this sample and the sample which precedes it is maximum. By way of example, Posi can correspond to an integer between 1 and N.

Figure 4 illustrates the OMaxi deviation determination principle. The curve SW represents, schematically, a part of the waveform of the signal SA. The instants tj, i, j being an integer varying from 1 to N and i being the index of the data packet PACKi, correspond to the successive instants of supply of a sample SAMPLE by the analog-digital converter 12. The packet of PACKi data is determined by the preprocessing circuit 14 from the N successive samples provided at times t1, i to tN, i. Oj, i, for j between 2 and N, is the difference or deviation between the value of the sample at time tj, i and the value of the sample at time tj_1, i. The difference O1, i corresponds to the difference between the value of the sample at time t1, j and the value of the sample at time tN, j-1 • The difference OMaxi corresponds to the difference Oj, i, for j varying from 1 to 25 N, which is maximal and is thus obtained as follows: N AMax = MAX (A, i) i = 1 where MAX is the maximum function. Posi position data can then correspond to the index j for which the difference Oj, i is maximum.

Referring again to Fig. 2, in step 46, whenever the preprocessing circuit 14 has determined a PACKi data packet, it is stored at a particular location in the memory 18 determined by the microphone. -processor 16.

In step 48, the microprocessor 16 successively links the PACKi data packets stored in the memory 18 and performs the following operations on each PACKi data packet read. The microprocessor 16 extracts the identifier Idi and the group of Sound-Datai data from which it forms a digital file, called a sound file SF, which is stored in the memory 22 at step 50. In addition, the microprocessor 16 processor 16 extracts from the data packet PACKi, the data group DATAi (including the difference AMaxi and the position data Posi) which is stored in the memory 20 in step 52. FIG. 5 schematically illustrates an example The memory 20 is, for example, a first-in, first-out memory comprising a number K of "boxes", each box corresponding to a sufficient memory space for the storage of a group of cells. DATAi data. Assuming that the last packet processed by the microprocessor 16 has the index M, the memory 20 allows the storage of K data groups DATAi, with j varying from MK to M. The data group DATAM_K, respectively DATAM, is therefore the oldest data group, respectively the most recent, stored in the memory 20. Each data group DATAi being associated with a data packet PACKi, which itself corresponds to N samples of the signal SD received at the frequency of sampling F, the memory 20 allows the storage of data relating to the signal SA over a duration of K * N / F seconds. By way of example, such a duration can be of the order of 6 seconds. The capacity of the memory 20, that is to say the number K of cells of the memory 20, can be parameterized by the operator. Figure 6 illustrates the second steps of the present exemplary method of detecting synchronization marks. In step 60, an operator operates the switch 26, which is detected by the microprocessor 16. In step 62, the microprocessor 16 controls the preprocessing circuit 14 so that the next PACKM data packet that the circuit In addition to the IdM identifier and the data groups DATAM and Sound-Datant, preprocessing 14 will provide a marker indicating that the switch 26 has been actuated. As a result, when in step 48 described above with reference to FIG. 2, the microprocessor 16 reads in the memory 18 the PACKM data packet containing the marker and stores in the memory 20 the data group. DATAM, it determines, in addition, in step 64 a datum AMaxMax which corresponds to the maximum deviation among the deviations AMaxM_K to AMaxM stored in the memory 20. The maximum deviation AMaxMax is thus obtained according to the following relation: M AMaxMax = Max (AMax) j = MK Assuming, for example, that AMaxMax is the difference AMaxL, L being between MK and M, the microprocessor 16 determines, from the position data PosL associated with the group of DATAL data and IDL identifier of the PACKL data packet, a REFT time reference representative of the time elapsed between the start of the recording of the audio sequence and the time when the maximum deviation AMaxL was obtained. By way of example, the time reference can be obtained in the following way: REFT = L * (N / F) + PosL / FA step 66, the microprocessor 16 stores in the memory 22, in an associated data file to the sound files 25 of the recorded audio sequence, the following data: an identifier REFC of the synchronization mark which has been detected; the time reference REFT; and the maximum deviation AMaxMax that has been determined.

The identifier REFC corresponds, for example, to a letter of the alphabet "A", "B", "Cu, etc. The use of a REFC identifier allows an operator to actuate the switch 26 several times. during the recording of a single audio sequence, because at each detected synchronization mark there is associated a different REFC identifier, which makes it possible to distinguish the synchronization marks during the subsequent editing step. In step 68, the microprocessor 16 controls the display on the screen 24 of information to remind the operator that he has commanded the detection of a synchronization mark, and to indicate him the identifier REFC The quality factor corresponds, for example, to the ratio between the value AMaxMax and a percentage of the synchronization mark determined. digital signal excursion Several audio tracks can be recorded simultaneously by the recording system 5. Indeed, the same audio sequence is, most often, recorded on several sound tracks from several microphones. For example, for a dialogue between two characters, each character may carry a microphone (usually concealed), for example, a high frequency link microphone connected to the recording system 5. Generally, a third sound track is recorded by means of a room microphone, for example, suspended at the end of a pole. As a result, the preprocessing circuit 14 may receive a plurality of digital signals SD associated with the different sound tracks. It can then be provided that the operator selects, before the start of the recording, the sound track on which the detection of synchronization marks must be performed. This may be advantageous in that one of the sound tracks is generally more suitable than the others for the detection of synchronization marks. This is, for example, the sound track obtained by means of the room microphone. The Sound-Datai data group of a data packet PACKi can then include the N samples associated with each sound track while the DATAi data group is associated with the sound track selected by the operator and is obtained as this has been described previously. As a variant, it can be provided that the operator selects all the sound tracks for the detection of the synchronization marks. In this case, for a data packet PACKi, the preprocessing circuit 14 determines, for each sound track, the difference AMaxi as described above and retains only the maximum of the differences AMaxi to provide the data group DATAi .

During the recording of an audio sequence, the microprocessor 16 may keep in a supplementary memory, not shown, the maximum deviation among the differences AMaxi determined throughout the recording of the audio sequence. To do this, each time the microprocessor 16 reads a packet PACKi data in the memory 18, it can determine if the difference AMaxi is greater than the last stored distance and, if so, stores the data Idi, AMaxi and Posi in the extra memory. At the end of the recording of the audio sequence, if the operator has not operated the switch 26 during the recording, the microprocessor 16 can determine the time reference REFT associated with the difference AMax1 stored in the additional memory and can store, in the memory 22, in a data file associated with the sound files of the recorded audio sequence, the difference AMaxi stored in the supplementary memory and the time reference REFT. During the subsequent editing step, the operator can read directly from the data file associated with the sound files of the recorded audio sequence the time positions of each synchronization mark. It no longer has to manually find the synchronization marks. In the embodiment previously described, a synchronization mark is determined from the difference AMaxi which corresponds to the difference between the value of a current sample and the preceding sample, which corresponds to a approximate value (divided by the sampling frequency F) of the derivative of the signal SA for the current sample. Such a calculation has the advantage of being particularly simple to implement at the level of the pretreatment circuit 14 which is generally also highly stressed. However, if the calculation capabilities of the preprocessing circuit 14 allow it, the AMaxi data can be obtained by more complex algorithms. Furthermore, in the embodiment previously described, the deviations AMaxi are determined as and when the preprocessing circuit 14 receives samples of the signal SD. When the operator actuates the switch 26, the microprocessor 16 determines the maximum deviation AMaxAMax from the K AMaxi deviations stored in the memory 20. This is a very simple calculation that can be done very quickly by the microprocessor 16 without slowing it elsewhere. However, if the computing capabilities of the microprocessor 16 permit, it is clear that the deviations AMaxi (and the maximum deviation AMaxAMax) could be determined only after the operator has actuated the switch 26. In this case, the memory 20 is no longer present and the microprocessor 16 uses the data stored in the hard disk 22 to determine the deviations AMaxi. Of course, the present invention is susceptible to various variations and modifications which will be apparent to those skilled in the art.

Claims (10)

  A method for detecting at least one synchronization mark of a digitized audio sequence, comprising receiving first successive digital audio data (SD) representative of the digitized audio sequence and characterized in that it comprises the following steps: (a) detecting the operation of a control means (26) by an operator during the reception of the first data; (b) setting a time window in relation to said detection; and (c) determining, from the first data received during said time window, a second datum representative of an instant of occurrence of the synchronization mark during said time window. 15   A method according to claim 1, of storing third digital audio data (SF) obtained from the first data and forming the digitized audio sequence and storing the second data with the third data.   3. Method according to claim 1 or 2, wherein the detection of the actuation of the control means (26) corresponds to the end of the time window.   4. Method according to any one of the preceding claims, further comprising the following steps: (d) determining, for each successive set of a first number of first received data, a fourth datum (AMaxi) representative of the maximum variation of the first data of said set and a fifth data item (Posi) representative of the instant at which said variation is maximum; and (e) storing, for a second number of the last received sets, the second number of fourth associated data and the second number of fifth associated data, and wherein, in step (c), the second datum corresponds to the fifth datum associated with the fourth maximum datum among the second number of fourth data.   5. The method according to claim 4, wherein step (d) consists in determining, for each set, the difference (Di) between each first datum of said set and the first preceding datum, the corresponding fourth datum (AMaxi). at the maximum distance.   6. System for detecting at least one synchronization mark of a digitized audio sequence, comprising means for receiving (14) first successive digital audio data (SD) representative of the digitized audio sequence, characterized in that it comprises: control means (26) operable by an operator during reception of the first data; and determining means (14, 16), from the first data received during a fixed time window in relation to the actuation of the control means, of a second datum representative of an instant of occurrence of the synchronization mark. during said time window.   The system of claim 6, further comprising: means (16) for providing third digital audio data (SF) obtained from the first data and forming the digitized audio sequence; and first means (22) for storing the third data and the second data.   The system of claim 6 or 7, further comprising: means (14) for determining, for each successive set of a first number of first received data, a fourth datum (AMax) representative of the maximum variation first data of said set and a fifth data item (Posi) representative of the instant at which said variation is maximum; and a second storage means (20), for a second number of the last received sets, the second number of fourth associated data and the second number of fifth associated data, and wherein, the second data determination means determines the fifth data associated with the fourth maximum data among the second number of fourth data.   9. System according to claim 8, wherein the means (16) for determining the fourth datum (AMaxi) is adapted to determine, for each said set, the difference (Ai) between each first datum of said set and the first datum. previous, the fourth datum (AMaxi) corresponding to the maximum deviation.   The system of claim 9, wherein the second storage means (20) is a first-in, first-out memory.