MXPA97009127A - Da's diffusion system - Google Patents

Abstract

The present invention relates to a data dissemination system (D), that information is transmitted in the pass band of a broadcast audio signal (S), characterized in that it includes the means of determination in at least one frequency band (S). F'13, ..., F'24) of the amplitude (A'13, ..., A'24) of the radiofrequency signal (S) and in comparison of this amplitude with a level of auditory concealment Nm ( 13), ..., Nm (24)) associated with this frequency band, the means of eliminating the frequency components of the audio-frequency signal in said frequency band if the amplitude of the signal is lower than the level of auditory concealment. of said band, and means for inserting said data in this frequency band at a level less than or equal to the level of auditory concealment of said frequency band.

Description

SYSTEM OF DIFFUSION OF DATA.

The invention relates to the field of signal diffusion comprising an audio-frequency component. More particularly, it refers to a data dissemination system.

The field of broadcasting (broadcasting of television or radio programs, wireless telephony, etc.) is well known.

A current trend is to issue, in addition to the programs (or the voice in the field of telephony), useful data for broadcasting societies, for control bodies, or by auditors or viewers. These data could refer for example: the help of the selection of a program of sound or television broadcast (example: help to automatic agreement, search by name of a radio station, search by type of program, search on demand, etc.). information on the program being broadcast or played after the recording (for example the name of the company that created a program, the title of the film broadcast by a radio station, etc.), REF: 26030 -, the service data in the case of analogue radiotelephony.

We also assist in the development of interactive dissemination systems that allow viewers or auditors to dialogue more or less perfectly with the origin of the program. These means are used either to act on the content of the broadcast program, whether to play and to bet, to communicate to the subject of that same program. Thus, recently a form of interactivity has appeared through small devices, simulating a pseudo-dialogue with a program designed for this purpose. A remote-controlled box gives the illusion of interactivity to the extent that it allows, for example, responding to a televised game of questions and / or answers as questions are asked. Or better yet, an electronic device concealed in a stuffed toy allows the latter to react to a program broadcast or reproduced with the help of a video cassette player. In fact, interactivity is not real because the series of good responses or the reactions of the toy obey the pre-established sequences, common to the memory of the interactive device or to the broadcast or reproduced program. The audiovisual sequence having been pre-registered according to a chosen code, its development is foreseeable and consequently, the only information to be transmitted al, interactive device is a starting signal so the exact score of questions / answers or the various possible reactions in the case of a toy.

There is also a request that refers to the automatic identification of a sound sequence, accompanied by an image or not. For broadcasters, it is a question of verifying that a given program is being broadcast well on the sequence assigned to it, which can become quite complex when a national program is affected by regional or local ruptures. This also allows, for the verification bodies, to account for the dissemination of works protected by copyright or to verify the conformity of the diffusion of advertisements. Finally, for the organizations of sounding or audience evaluation, it is a question of quickly identifying what is really heard by an auditor or a viewer. Currently, to evaluate the audience of radio receivers, the only solution available is the consumer interview survey.

All these concepts are easy to introduce when designing new broadcasting systems, particularly numerical, radio or television. On the other hand, the systems and parks of existing materials generally lend themselves poorly to this evolution and experience proves that a technical-commercial point of view, the compatibility and the relative cost to the procedures and devices to be used are determining factors in the introduction of new services.

For the emission of data that refer to a broadcast program, these techniques are currently used: A first technique consists of transmitting this data outside the busy band, occupied by the signal of the program (sound and eventually picture) transmitted. One solution consists, for example, in sound diffusion by "multiplex" frequency modulation, or in using the upper part of the "multiplex" between 54 and 76 ilohertz. Another example is to use the lines available during the return of the section in television broadcasting. These techniques have drawbacks.

The saturation of frequency resources available in diffusion limits the number of users of those resources. On the other hand, it is necessary to have receivers adapted to the concurrent bands used to transmit the information issued.

Another technique consists of transmitting the data in the concurrent band of the transmitted program signal, which does not need the use of dedicated frequency bands. Do not It is therefore necessary to use transmitters and receivers that are adapted to a transmission in similar frequency bands. Typically the source signal (corresponding to the program to be transmitted) is filtered in order to eliminate the frequency components in a given frequency band and the data is included in this band. The signal of origin is therefore deformed, which can be annoying for a viewer or an auditor to whom the data does not interest him. Consequently, the time devoted to sending information is limited by the broadcasters to the strict minimum, which reduces both the performance of the data. Thus, in the framework of interactive devices in the television domain, the data load is made globally, in a single time, at the beginning of a given application. It is not then possible to adapt the data in response to a modification of the program, which must be developed according to the expected schedule and without unexpected interruption. It is of course possible to use the filtering means at the receiver level so as not to systematically reflect the received data from a sound or visual point of view, this being transparent for the auditor or the viewer. However, it can be assured that the signal seen or understood by the viewer or the auditor will be identical to the original signal that he had perceived, before the insertion of the data.

In view of the foregoing, the objective of the invention is to propose a system that allows data to be transmitted in a concurrent band of a signal comprising an audio-frequency component, without modifying, in comparison to the original audio-frequency signal, the signal perceived by the auditor The invention proposes to insert these datps in the said, hidden frequency bands of the original audio-frequency signal, if those bands exist, ie at a level below the level of instantaneous hearing due to the phenomenon of auditory concealment induced by the same original audio-frequency signal . The transmitted data are then inaudible, without altering the audio-frequency signal of origin from a subjective point of view, and without requiring the use of frequency components located outside the spectral band occupied by the source signal. The invention thus proposes a data transmission adapted to the use of existing receivers and transmitters, and subjectively non-disruptive to the auditor.

Thus, the invention relates to a data dissemination system, said information is transmitted in the pass band of a diffuse audio-frequency signal, characterized in that it comprises, as a means of determining, at least one frequency band of the amplitude of the audio-frequency signal and of comparison to this amplitude with a level of auditory concealment associated with this frequency band, the means for eliminating frequency components of the audio-frequency signal in said frequency band if the amplitude of the signal is lower than the level of auditory concealment of said band and the means for inserting said information in this frequency band. frequency band at a level less than or equal to the level of auditory concealment of said frequency band.

Other features and advantages will appear on reading the description that follows, to read together the attached drawings in which: Figures 1 and 2 represent the diagrams that illustrate the phenomenon of auditory concealment, Figure 3 represents a data extraction device, Figure 4 represents a data insertion device.

Figures 1 and 2 are the diagrams of amplitude in relation to the frequency illustrating the phenomenon of auditory concealment, which is a phenomenon of psychological origin.

If a human being is aware of a given frequency and amplitude frequency signal, the phenomenon of auditory concealment is translated by the non-perception, by that same human being, of audiofrequency signals emitted simultaneously and having the amplitudes lower than levels of given stages.

Thus, referring to Figure 1, if we consider a monofrequential signal of frequency fO, located in the audio-frequency spectrum (typically between 20 and 15 500 hertz) and the amplitude Ao, that can define an M domain (fO, AO) in amplitude and in frequency such that all monofrequential signal transmitted simultaneously, of frequency fs comprised in a limited frequency domain [fOm, fOM], with fOm < fO and fOM > fO, and amplitude A < A (fs, fO, AO) < AO is inaudible.

The values fOm, fOM are variable by a given frequency fO. Practically, more important is the amplitude AO, but the domain [fO, fOM] is broad. It will also be noted that the domain is not symmetric in relation to fO, and extends more widely in the domain of frequencies above fO.

The amplitude value A (fs, fO, AO) varies according to fs, fO and AO. Practically closer this fs of fO plus the level A (fs, fO, AO) of inaudibility is important.

The phenomenon of auditory concealment has been known for many years. For more precision we will refer to the work "Psicoacoustics, by E. Zwicker and R. Feldtkeller, Ed. Masson, 1981". The experimental results described in this work have led to a standardization (ISO / IEC 11172-3 standard).

We can define a curve of the level of concealment M (S) (illustrated by a dotted line in figure 2) by every S signal covering the audio-frequency spectrum [fm, fM], with fm = 20 Hertz and FM = 15 500 Hertz . In the example illustrated in figure 2, we will notice that there are two domains [flm, flM] and [f2m, f2M] in which the curve of the level of concealment M (S) has an amplitude greater than that of the signal S. Specifically , that means that the spectral components included in those domains are inaudible to the human being. Accordingly, the auditory return of a signal S 'identifies the signal S outside those domains, and without frequency components in those domains, it will be identical to the return of the signal S illustrated in figure 2.

The modeling of the phenomenon of auditory concealment has led to the division of the audio-frequency spectrum into twenty-four disjointed domains, called critical bands, such that the meeting of twenty-four critical bands covers the frequency domain between 20 hertz and 15 500 kilohertz. Each band e is defined by its central frequency, faith and its breadth.

The bottom panel gives for each critical band the value of the central frequency and its amplitude.

We will notice that the critical bands have variable amplitudes, the least wide is the critical band Bl, which covers the most serious frequencies and the widest is the twenty-fourth critical band B24 that covers the most acute frequencies.

For each critical band, ISO / IEC 11172-3 defines a level of critical band concealment Nm (i). This is an approximation of the level of the curve of the level of concealment over the whole of the critical band (the actual level of the curve of the level of concealment by a given signal may vary in the same critical band). The level of concealment Nm (i) is defined as a function of the levels of concealment of the eight lower critical bands (Nm (i-8) to Nm (il), if they exist, and of three higher bands (Nm (i + 1) to Nm (i + 3)), if they exist.

We have Nm (i) = S Nm (j), with j positive integer index such that je [i-8, ..., i-1, i + 1, ..., i + 3] Nm (j) = 10 [Xnm (j) - Av (j) - Vf (j)] / 20, Xnm (j) = 20 loglO (Av (j)) + 5.69 dB (acoustic pressure), Av (j) = 6.025 + 0.275 * z (j) for the tonal stripes, Av (j) = 2,025 + 0,175 * z (j) for the non-tonal stripes, with Av (j) the hiding index of the critical band ja jz (j) the rate of the JTSlipabanda criticizes, Vf (j) = (ij-1) * (17 - 0.15 * Xnm (j)) + 17, by j from i-8 to il, and Vf (i + 1) = 0.4 * Xnm (i + 18) + 6, Vf (i + 2) = 17 * Xnm (i + 2) + 6, Vf (i + 3) = 34 * Xnm (i + 3) + 6. z (j) is a constant defined by each critical band and we have z (l) = 0.62 dB, x (2) = 1.8 dB, z (3) = 21.4 dB, z (4) = 3.6 dB, z (5) = 4.7 dB, z (6) = 5.8 dB, z (7) = 6.7 dB, z (8) = 7.7 dB, z (9) = 8.9 dB, z (10) = 10.0 dB, z (ll) = 10.9 dB, z (12) = 12.0 dB, z (13) = 13.1 dB, z (14) = 14.0 dB, z (15) = 14.9 dB, z (16) = 15.8 dB, z (17) = 16.7 dB, z (18) = 17.7 dB, z ( 19) = 18.8 dB, z (20) = 19.8 dB, z (21) = 20.9 dB, z (22) = 22.2 dB, z (23) = 23.2 dB, and z (24) = 23.9 dB.

In general, the most hidden bands are the acute bands of the audio-frequency spectrum that are hidden by the serious, statistically more energetic bands.

After this brief assessment of the phenomenon of auditory concealment and its modeling, we will describe an example of the application of the invention that consists in transmitting the data in the pass band of a diffused audio-frequency signal.

The data may also be analog (musical drawings for example) that numerical (ie binary data). The Recording to the broadcasting system is widespread (for example the name of a radio station or the references of musical titles issued by this station) and have a vocation to be perceived by the auditor, for example by means of an announcement. of liquid crystals. They may also be the service data of interest to the signal diffuser or regulatory instances, and be imperceptible by the auditor.

In the sequence of the description given by way of example, we will assume that the data is binary data.

These data will be relative, for example, to the programs broadcast by a radio station.

A radio station usually broadcasts in the direction of these auditors the modulated audio-frequency signals by means of classical amplitude or frequency modulation techniques. The audio-frequency signals may be a song, a musical record, the voice of an animator, etc.

The invention proposes to calculate from the audio-frequency signal to emit, by one or more of the critical bands Bi of the audio-frequency spectrum, the level or levels of concealment of this or these critical bands. If for a band criticizes the level of concealment is higher than the level of the audio-frequency signal, we can eliminate, without noticeable difference by the auditor, the part that corresponds to the audiofrecuencial signal. The invention proposes to insert the data (we will speak of audio-frequency data signals), inaudible to the auditor, in this critical band, or a part of this critical band part, in the place of the original audio-frequency signal (hence, of course, the level of the audio-frequency data signal is lower than the level of concealment of the critical band). In reception of the transmitted signal, it is sufficient to filter the received signal according to the critical bands in order to separate the audio-frequency signal from the data and to process the transmitted data.

We will notice that the transmitted information capacity can not be made in fixed practice, the original signal (and thus the corresponding critical band concealment levels Nm (i)) being preferably variables in time, either in frequency or in amplitude.

A data transmission system according to the invention will mainly include a data device (whose example is illustrated in Figure 4) and a data reception device (the example of which is illustrated in Figure 3).

Typically, the data insertion device may be applied either to the state of a final sound or visual broadcast control, or to the state of the production of the audio-frequency signals. The data reception device will include, for example, a device for announcing received data (if the data is intended for the auditor) and / or a storage device (if the data is dedicated, for example, to a definite time audiometry control). The receiving device may also include a device for re-sending information, for example to a game box in the context of interactive television programs. The audio-frequency data signal can be collected, at the level of the reception device, either acoustically by a simple microphone (arranged near the radio receiver's horn), either electrically with the help of an appropriate connector (such that an output of audio recording).

In relation to figure 4, we will describe by way of example a data insertion device 1, that information is in the present case of the binary data.

To transmit the data in the audio-frequency signal of a radio or television program, we change the signal by a modulation in certain frequency bands of that signal. number ca. This transmission is preferably made at a level below the levels of concealment of these frequency bands, so as to ensure the inaudible nature of the information transmitted. On the other hand, this transmission is preferably done when the levels of concealment are sufficiently high to ensure a satisfactory signal-to-noise ratio in relation to the diffusion channel.

In an example, the data to be emitted can be organized in sections consisting of a start word and a defined number of data words. We can also choose a section that includes a word of principle, a variable number of data words and a word of end.

The data insertion device 1 illustrated in Figure 4 includes an input 2 for receiving the original S-audio signal to be broadcast (song, voice of an animator, etc.), an input 3 for receiving the data D to be broadcast, and a output 4 to provide an audio-frequency output signal S 'produced from the audio-frequency signal S of origin and data D.

The audio-frequency signal S is filtered in a bank of twelve band-pass filters FPB'13 to FPB'24, preferably complexes receiving the audio-frequency signal S on input. He Analytical treatment of the S signal facilitates the calculation of the amplitudes. Each complex filter outputs the real part (R'13 to R'24) and the imaginary part (I '13 to I'24) of the audio-frequency signal S in the frequency band (notices F'13 to F'24). ) that he lets go. As we will see, the bank of complex band-pass filters FPB'13 to FPB'24 allows to eliminate the components of the audio-frequency signal S in the frequency bands F'13 to F'24 to insert the data. These frequency bands (F'13 to F'24) are the bands included in the critical bands B13 to B24. A calculation instrument of the amplitude OAC1 calculates the amplitudes A'j (j integer index of 13 to 24) from the signals R'j and I'j provided by the filters FPB'13 to FPB'24.

The audio signal S is likewise filtered in a bank of twenty bandpass filters FPB5 to FPB24, preferably complexes, receiving the adiofrequential signal S on input. Each complex filter produces in output the real part (R5 to R24) and the imaginary part (113 to 124) of the audio-frequency signal S, in the frequency band that it passes. The bank of complex band-pass filters FPB5 to FPB24 allows calculating the levels of concealment of critical bands B13 to B24. This calculation is made from an organ of calculation of amplitude OAc2 calculating the amplitudes Ai (i entire index of 5 to 24) from the signals Ri and Ii provided by the FPB5 and FPB24 filters. These amplitudes are provided to an ON calculation processor by calculating the hiding levels Nm (13) to Nm (24).

The amplitudes A'13 to A'24 and the concealment levels Nm (13) to Nm (24) are provided to an OC command organ which will compare them two by two in order to determine if there are two amplitudes A ' jl and A'j2 lower than the corresponding concealment levels Nm (jl) and Nm (j2) (jl and j2 being two different integers between 13 and 24). If this is the case, there are at least two frequency bands F'jl and Fj2 in the audio-frequency spectrum by which the signal S is inaudible. It is then possible to filter the signal S in order to eliminate these spectral components in those two frequency bands F'jl and F'j2.

To do that, the real components, annotated R'l and R'2 of the signal S in those two frequency bands F'jl and F'j2, of the original signal S are subtracted. Those two components R'l and R'2 are provided by means of a multiplexing device MUXP receiving the components R'13 and R'24, each of these components are weighted in such a way that we cancel them all except two of them (R'jl and R'j2). The command of that MUXP device is made by the OC command organ.

These components (we have for example R'l = R'jl and R'2 = R'j2) are immediately subtracted from the signal S (having been delayed this to take into account the term of the crossing of the filters and the device multiplexing) in two adders SMl and SM2, in such a way that an audio-frequency signal S'M = S R'l-R'2 is produced. That audiofrecuencial signal S'M is subjectively identical, for an auditor who will perceive it, to the signal S.

The set formed by the band-pass filters F'13 to F'24, the multiplexing device MUXP and the adders SMl and SM2 behave as a short-band filter adapted with respect to the signal S.

The frequency bands F'jl and F'j2 are released to allow the insertion of the data D, we are now going to be interested in this insertion.

Classically, we will first proceed to a binary data form D. We will note that the realization of this fitness is independent in every state of cause of the release of frequency bands F'j in the audifrecuencial signal S. The data D to transmit are placed in the form of the desired section (that is, inserting the words of principle and, eventually, end, the redundant codes etc.). Then, we will produce two audio signals of Si and S2 data by means of a MOD modulator. The numerical modulation used will be, for example, a QPSK (Quadrature Phase Shift Keying) modulation, the data in shape, encrypted in NRZ (without return to Zero) by modulating in phase two frequency carriers included in the bands F'jl and F'j2 , preferably corresponding to the central frequencies of the bands F'jl and F'j2 used (which allows the full range of these bands to be used to emit the audio-frequency signals of SI and S2 data). This modulation step requires, of course, the knowledge, by means of the command organ OC, of the frequency bands released in the spectrum of the signal S.

Parallel to the release of the bands F'jl and F'j2, the levels of concealment Nm (13) to Nm (24) are provided by the ON organ to a MUXN multipressing device that will output two levels N '. m = Nm (jl) and N "mONm (j2) In order to take into account the modulation chosen to produce the signals SI and S2, we produce two coefficients N 'and N" from the coefficients N'm and N "m, with the help of a CAG gain control device." With the help of two multipliers Ml and M2, we immediately produce two audio-frequency data signals S'1 = N '* S1 and S2 = N "* S2: Adding in two addiners SM3 and SM4, the signals S'l, S'2 and S'm, produce a signal S '= S- (R '1 + R' 2) i- (S '1 + S' 2). The signal S 'produced includes both the audio-frequency components audible to the original audio-frequency signal S and the data D (represented by S'l and S'2 which are inaudible.

Once the signal S produced, we will classically modulate it according to the known techniques before emitting it in the direction of the receivers of the auditors.

We will notice that the gain applied to the signals SI and S2 being only proportional to the levels of concealment of the jl-ava and j2-ava bands F'jl and Fj '2, the amplitude level of the signals S'l and S' 2 may be greater than the amplitude levels of the components of the S signal that have been raised.

Preferably, the bands F'13 to F'24 have an equal amplitude to ensure a performance of the transmitted data that is fixed, whatever the bands F'13 to F'24 used to transmit them. We can thus use the same type of modulation, regardless of the bands released in the signal S. In the illustrated example, we foresee the possibility of transmitting the data in the last twelve critical bands, of the critical band B13 (fc = l 860 Hz ) to critical band B24 (fc = 13 750 Hz). As we have seen, this information is transmitted in two bands each one in one of the twelve critical bands. Of course, more important is the number of bands F'j used simultaneously, the higher the performance of transmitted data. We can then make a data insertion device using all the releasable F'j bands. However, we will note that the simultaneous use of a small number of bands F'j allows to reduce the probability of distortion of the original audio-frequency signal if it varies considerably from one instant to the next (although this probability is slight considering the temporary concealment of the human ear).

Whatever the critical bands in which we insert the data, we will clearly understand that the band (s) F'j used within those critical bands have an amplitude less than or equal to the corresponding critical band amplitudes.

In the illustrated example, the first bandpass filter bank F'l3 to F'24 of amplitudes equal to 280 Hz at -3 decibels. This amplitude corresponds to the amplitude of the critical band used to insert the data that have the amplitude the lightest, that is, to the amplitude of the thirteenth critical band (of course, we assume here that the carrier frequencies used to produce the signals audio frequencies are equal to the central frequencies of the critical bands). We therefore have little interest in starting a data transmission in the lower critical bands, they are having a lighter amplitude, which would limit the maximum admitted performance.

The filter bank F'13 to F'24 is preferably made by multicance filtering, which allows having a constant propagation time and a limited number of operations.

The second filter bank F5 to F24 is preferably obtained from reconstructible bandpass filters (ie filters such that the sum of the filtered signals at output is identical to the input signal before filtering) whose sizes correspond to the critical bands. In other words, we are interested in calculating the levels of concealment of the critical bands as precisely as possible, which avoids producing the audiofrequency signals of data that could be audible.

The binary information is for example grouped into thirty-two bit words. A transmitted section will consist for example of a word of principle, coded in thirty-two bits and a data word. The word of principle is for example composed of the first nine bits that constitute a hooking ramp used in the receiving device, the next twenty-three bits forming a synchronization word. The data word is for example composed of three octets that represent the data of a last redundancy octet for the implementation of an error correction code if such a code is used. This organization of sections of information corresponds to a transmission of the information on the temporary blocks of the audio-frequency signal of a duration of 256 milliseconds, which corresponds to the duration necessary to transmit sixty-four bits, that is, two data frames. We can thus achieve a maximum binary performance of 500 bits per second.

Preferably, the data sections are emitted provided that the levels of concealment of critical bands used to insert the data are higher than the minimum energy level that allows to resist the disturbances given by the channel.

Although it has not been specified, it is of course preferable not to release the frequency bands in the original audio-frequency signal when we do not have data to be transmitted. For this, it is enough to cancel the signals produced at the output of the MUXP multiplexing device. Thus, even if the level of concealment of the original signal varies rapidly and in a important, there will be no risk of disturbance of the original signal by suppression of audible frequencies. Once the data transmission is carried out, we will preferably proceed to a progressive cancellation of the output signals of the multiplexing device MUXP in order to reduce the probability of returning the audible "delay".

If the level of concealment of the original audio-frequency signal decreases and we have issued the word of principle, we will preferably continue to broadcast in order to facilitate the processing of the data at the receiving device level. If the data is encrypted in thirty-two bits it is little disturbing of the fact of temporary auditory occultation.

The data extraction device 5 illustrated in Figure 3 includes an input 6 for receiving the frequency audio signal S '. The audio signal S 'is filtered in a bank of twelve filters FPB' 13 to FPB '24 having identical patterns with the twelve FPB' filters, to FPB '. Thus, twelve audio-frequency signals 13 24-S'13 to S'24 are produced, which correspond to the spectral components of the signal S 'in the bands t "i3 to' 24 in which the data inserted by a device are susceptible. analogous to that described with reference to figure 4. Device 5 includes a bank of twelve DEMOD13 demodulators DEMOD24, each demiler is associated with one of the band-pass filters. Once the demodulated signals, we will prepare them in a sample EC13 to EC24 associated with the clock recovery devices RC13 to RC24, in order to produce the binary data.

Once the audiofrecuenciales signals put in samples, the produced binary data are treated in the organs of recognition RTB13 to RTB24 with the purpose of to determine if those data are representative of the data transmitted (in the case that the bits of synchronization of word of principle are present) or if those data do not correspond to anything (the probability being quite slight that we can produce by sample, from any audio-frequency signal, the bits corresponding to the synchronization bits of a word of principle).

Of course, if the data broadcast is not numeric data but analog data, such as a musical reason for example, we will adapt the insertion and data extraction devices accordingly. In particular, it will not be necessary to use the modulation, modulation and sampler devices. These will be replaced by the means of frequency transposition of the data to the frequencies released in the insertion device.

Claims (8)

1. CLAIMS 1- Data dissemination system (d) f these data are transmitted in the pass band of an audiofrecuencial signal (S) diffused, characterized in what includes: - determining means in at least one frequency band (F '13, ..., F' 24) of an amplitude (A '13, ..., A' 24) of the audio-frequency signal (S) and the comparison of this amplitude with a level of auditory concealment (Nm (13), ..., Nm (24) associated with this frequency band, - the means for eliminating the frequency components of the audio-frequency signal in said frequency band if the amplitude of the signal is lower than the level of auditory concealment of said band, and - means for inserting said data in this frequency band at a level less than or equal to the level of auditory concealment of said frequency band. 2- System according to claim 1, characterized in that the frequency band (F '13, ..., F' 24) is comprised in a reference band (B13, ..., B24). 3, - System according to claim 2, characterized in that the frequency band has a central frequency of the reference band. System according to claim 1, characterized in that it comprises the means for inserting the data into at least two different frequency bands (F'jl, F'j2). 5- System according to claim 4, characterized in that the bands of different frequency are included in the reference bands (Bjl, Bj2) of different amplitudes. 6- System according to claim 5, characterized in the bands of different frequency have a central frequency identical to the central frequency of the reference band in which they are comprised. 7- System according to one of claims 5 to 6, characterized in that the bands of different frequency have the same amplitude. 8. System according to claim 7, characterized in that the amplitude of the bands of different frequency is equal to the amplitude of the reference band in which they are comprised having the amplitude the lightest.