FR2645692A1 - System for the secret transmission of audio signals: cryptophony by spectral inversion. Television, radio, telephone according to the process. Devices for producing coders or decoders - Google Patents

Abstract

Coding process for the secret transmission of audio signals. The invention is a process of cryptophony with spectral inversion and variable turnover frequency in discrete, abrupt and rapid steps. It consists of a coder using a modulator 3 with variable frequency 5 preceded by a filter 2 and followed by a filter 4 whose cutoff frequency can vary according to 5. The variable frequency 5 is generated by a programmable divider 54 of a reference frequency 56. A decoder using a modulator 10 with variable frequency 12 is preceded by a low-pass filter 9 with switched capacitances whose central frequency can vary as a function of the modulating frequency 12. The device according to the invention is particularly intended for coding the sound channel of a radio chain or pay-television chain, and for audio coding for radio telephones.

Description

SYSTEM FOR THE SECRET TRANSMISSION OF AUDIT SIGNALS
The present invention relates to cryptophony.

 The system uses a device for coding speech, music or any signal carried by an audio frequency channel passing through a transmission or storage means which could be intercepted by unauthorized users in such a way that those they cannot fully enjoy the audio-frequency signal. A decoding device allows the authorized user to take full advantage of the sound or musical message, or of the data transmitted or stored on an intermediate memory medium.

 The principle of the invention achieves temporally variable encryption with one or more scrambling levels which give the possibility of dynamically coding the sound signal. One of the characteristic aspects of the invention is the possibility of transmitting the musical signal in high fidelity without introducing parasites, bandwidth or dynamic changes due to the encryption process. This aspect is particularly interesting in particular for the transmission of pay television audio signals where the subscriber is entitled to have a quality signal and where the pay chain wants to protect itself against the manufacture and use of pirate decoders. If the signal encrypted by the pay-TV channel passes by a mass radio broadcast rather than by a cable transmission, the ease of piracy poses a crucial encryption problem for the chain which can question its credibility.

 The coding device used for the invention comprises: -an input low-pass filter intended to precisely limit the bandwidth of the audio signal, -a modulator intended to effect an inversion of the bandwidth of the signal around the modulation frequency, itself located above the bandwidth of the audio signal, and a low-pass filter intended to extract from the signal leaving the modulator, the components located in the low inversion band.

 The decoding device comprises a device which is either symmetrical or complementary to the coding system comprising in any case: a first low-pass filter intended either to limit the bandwidth received on the decoder, or to eliminate unwanted signals such as line frequency in TV, or to eliminate signals used elsewhere for signaling but which are not coded audio signals -a modulator operating at the same inversion frequency as that of the coder, -a second low-pass filter intended for extract from the outgoing signal of the modulator, the components located in the low inversion band.

 The principle of inversion by audiofrequency spectrum modulator around a fixed carrier during transmission has been known for many years; in fact, the first French patent of spectrum inversion cryptophony N 1,285,941 dates from 1961.

In all inverting spectrum cryptophony methods, the main problem is to obtain, with great economy of means, encryption which varies over time and which does not cause unwanted clicks or limitation of bandwidth. A system intended for example for coding the encrypted channel Canal Plus', patent N "2.580.885, successively performs a double spectrum inversion firstly around a fixed frequency and then around a variable frequency between two values multiple of the other. To allow this instantaneous and abrupt variation of modulation frequency, this realization must obligatorily switch two frequencies multiple of the other otherwise, at the time of the transition, a transient is generated which s 'hears decoding.

 The invention described here proposes, not only to achieve an abrupt variation in modulation frequency, but also to provide a plurality of reversal frequencies which makes it possible to encrypt between, no longer two, but several dynamic frequencies of reversal.

 Thus, instead of having, as in the prior art, either an all-or-nothing encryption system, or a two-state encryption system using a double frequency inversion on transmission as on reception, we now have a system with several discrete turning frequency values. The prior art indicated, in particular in patent No. 2,416,597, means for making the modulation frequency variable by analog circuits.

 The present invention consists of an encryption system based on an inversion with respect to a frequency obtained by variable division ratios of a master clock, preferably a clock using a quartz resonator. These abrupt changes in modulation frequency take place at synchronous times between the transmitter and the receiver. One of the novel features of the invention is that the abrupt and instantaneous change in modulation frequency is not heard in the audio signal, if and only if the change in frequency, therefore in modulation period, takes place at the end of said entire period preceding the change.

 In other words, if we chain successively known successive periods of modulation, we can vary the shift of the resulting spectrum abruptly and very quickly, without causing audible distortions on the audio signal. By abrupt, we mean a spectrum jump of the order of, for example 1000 Hz instantaneously, but not limited to this value. The difficulty linked to the invention is to synchronize within reasonable limits the instants of change of modulation period on transmission and on reception, but this system does not impose any constraint on the moment of switching with respect to the modulated signal. , such as switching during silence.

 On the other hand, this cryptophonic system can use either a wired pseudo-random sequence generator to chain the sequences of modulation frequency values, or use to calculate this sequence a pseudo-random register calculation algorithm in maximum period, with or without complementary polynomial and with or without loopback matrix which can favorably be produced by a programmed microprocessor. The proper aspect of the generation of pseudo-random values is not dealt with in this description because we have focused on the means representing the encryption support, it being understood that it is assumed that the pseudo-random values deciding on the sequence encryption are generated according to the same algorithm on transmission as on reception and that the means for generating pseudo-random sequences are known to those skilled in the art.

 We will describe below, means of synchronizing the moments of change of modulation period 9 transmission, as at reception.

 We will see later how the sequential change of modulation periods is carried out precisely at the end of an entire period, but an essential aspect of the invention is that the programmable divider which generates the modulation periods by account is loaded not directly by a value that can be modified at will at any time, but that the value to be deducted is continuously loaded or reloaded exactly at the end of a whole period or half-period. This system is advantageously carried out by a register located upstream of the up-down counter which contains the next value of period to be counted which will take effect only when the down counting reaches zero. In this way, and contrary to the systems known at present, we chain without jumps random phase, the moments of frequency change, which is an essential feature not to generate noise or clicks in the transmission. On the other hand, the moments of modulation frequency change can be prepared in advance by a wired auxiliary circuit or a programmed microprocessor, and loaded into the intermediate register at an asynchronous moment with respect to the modulation period, which facilitates greatly the wiring of the system upstream and in the case where this system is integrated in a microprocessor, this asynchronous loading does not oblige the software to either wait continuously for the right moment of loading, or to generate an interruption and serve very quickly the loading of this new value.

 Another advantage linked to the invention is the flexibility of the encryption depth. Indeed, by making the encryption frequency dynamic, we obtain two new parameters which qualify the encryption levels.

Let FO be the lowest frequency of modulation achieved, and F (t) the instantaneous value of the frequency at time t. The amount of frequency offset D = F (t) -FO, is the first parameter of the encryption level. The time between two frequency changes is the second parameter.

 The psychoacoustic effect of a variation in modulation frequency is appreciated with an encoder using variable modulation according to the circuits of the invention described below, and a fixed frequency decoder located at the frequency FO, simulating a pirate attempting to listen to the signal.

 Indeed, the shift D between the two frequencies used, causes a shift of the same value, constant on all the incident frequencies.

For example, if the incident signal F1 is sinusoidal at a frequency of 100
Hz, and the offset D of 7 Hertz, the sinusoidal signal generated FG is 107
Hertz, that is FG = FI + D. If Fl is a signal comprising harmonics such that F (1) = FI, F (2) = 2xFl, ..., F (n) = nxFI, the signal generated will comprise the same harmonics, at the same amplitude but such that FG (n) = nxFI + D. For example, if F (1) = 100, F (2) = 200, F (3) = 300, then Fi (1) = 107, FG (2) = 207,
FG (3) = 307. We note that, from a harmonic spectrum, we obtain an inharmonic spectrum because FG (2) is different from 2xFG (1) and
FG (3) is different from 3xFG (1). This inharmonicity is added to the overall frequency shift upwards (or downwards if we modulate with
D = FO-F (t)) to produce a harmful effect for two reasons on the intelligibility of the signal.

We note that an offset D of 5 to 10 Hertz of the spectrum is generally not detected by an untrained ear, but that an attentive professional musician will succeed in deciphering the inharmonicity of a musical signal passing through this transformation. note that if D is between 20 and 40, any listener will find the music "false". Even this small level of variable encryption can be sufficient to make speech intelligible (intentionally), but makes listening to music very annoying, and In any case, a recording of Hi-Fi quality is prohibited. If D is around 400, the listener still understands speech and certain noises, but the distortion makes it difficult to follow a dramatic action or a discussion, the music is inaudible. If D is around 1000, speech understanding is difficult, the gene constant. If D = 150O, we reach the practical limit of speech recognition, the sense of speech cannot be perceived. If D = 2000, only s some random words can be recognized. We thus have a reliable means of knowing the level of encryption which we want to apply for a transmission according to the quality and the infidelity which we want to transmit to a pirate. This first parameter D, will be according to the subject of the invention continuously variable in a pseudo-random manner, so that we cannot follow the level of encryption, even if we have the same reception circuit as that decided by the The effects mentioned above are guaranteed by changing for example every second the modulation frequency, which prevents an illicit decryptor from 'searching' for the level by trying all levels until a clear signal is obtained. It is to highlight that. these psychoacoustic tests are obtained under rigorous quality conditions, using as a signal source for example, a musical signal coming from a compact disc. On the other hand, we consider as 'high fidelity' a signal which would follow variable coding-decoding without an attentive musical ear being able to discern the slightest parasitic modulation or the slightest switching instant. We have, without any problem, the possibility of changing level much faster, for example from 5 to 20 times per second, without causing a perceptible loss of signal quality.

Another notable characteristic of the invention is the new principle of variable filtering as a function of the variation of the modulation frequency. In patent No. 2,580,885 already cited, we have page 8 lines 33 to 36 and page 9, lines 1 to 4, a finding of helplessness in the face of the variable frequency modulation problem. In one of the aspects of invention, if we realize, thanks for example to the use of low-pass filters with switched capacities (FCC for short), a precise, programmable and reproducible shift of the filtering frequency, we are sure to always have a constant bandwidth. In fact, the FCCs have their central frequency dependent on an external digital clock which can advantageously be generated by an oscillator controlled in phase by the modulation frequency. Another aspect of the FCCs, unknown to those skilled in the art, is that we can very quickly change the switching frequency without causing changes to the audio signal passing through the filter. This is on two conditions: the switching signal must change after a whole period otherwise we have s a switching transient, and the audio signal must obviously be bounded in frequency by the minimum frequency that the FCC can transmit, otherwise we perceive a variable low-pass filtering of the harmonics of the audio signal, which is normal because we change the frequency filter center. This possibility which has not been used, to our knowledge in a circuit, makes it possible to follow almost immediately the jumps in frequency of modulation that we impose, for example less than 5 ms for a jump of 1000 Hertz at
F0 = 13000 Hz. In the case where we control the FCC by a phase-locked loop (PLL in English), the servo-mechanism for hooking the loop is dependent on the time constant of the filter located downstream of the detector phase. It will be necessary to take care not to leave the possibilities of the hooking of the loop and to have a speed of response to the transients compatible with the frequency hopping that we impose. But two combinable possibilities are offered to us: we can limit the instantaneous jumps of frequency to a ceiling value, and realize a low-pass filter of PLL with weak time constant. The latter will cause twists exceeding the new target frequency value if the servo-mechanism is under-damped, but in this circuit configuration, it does not matter because the low-pass filtering of the FCC will integrate this exceeding of the switching frequency. We will see later in the description of the circuits, the specific orders of the filters that we have successfully used. If the use of FCC is very economical and allows thanks to the use of integrated technologies, a very large circuit compactness, it is quite possible, in particular to obtain a very high quality in the encoder located at the head of the network, to use low-pass filters whose cut-off frequency can be controlled by analog means, possibly controlled by a Digital Converter
Analog (DAC in English) such as the circuits commonly used in filters of analog music synthesizers which are variable state filters (State Variable Filters in English), programmable by a voltage, or alternatively switch by analog switch, capacitors or resistances representative of the cutoff frequencies of said filters that we want to select.

 Another notable characteristic of the invention is the use of a modulator by analog switches controlled by the modulation frequency. As we will see in the description of the circuit, the characteristics of rejection of the carrier frequency in the output signal are very good, because if we manage to pair the resistors two by two constituting it, we can obtain a rejection better than 60dB, which allows us to completely include the modulator function in an integrated circuit, without having to provide for external adjustment to the circuit, in particular to dose a reinjection of the carrier in phase opposition.

 To summarize, the circuits used in the context of the invention are simple and easy to integrate into a monolithic circuit, the levels of complexity of the system are variable without altering the principles of the invention to allow a type of cryptophony adapted to the needs of the application of secrecy.

For example, and without limiting the uses to these descriptions, we envisage a use of these circuits for the sound channel of a pay television; in this case, the encryption levels do not have to be very numerous (for example N = 16), the times between two rapid shifts (for example
T = 1 second) and large values (for example Dmax = 1000 0 Hz). It is a question of preventing a pirate decoder from comfortably following a transmission of
Pay TV without necessarily wanting to prevent the accidental understanding of a few words. In the case of coding for a pay radio
FM for example musical, just operate with N = 10, Dmin = 50, Dmax = 500
Hz, T = 20 s. We know that even a slight change in the musical signal makes listening very annoying, we are playing here on the disappearance of high fidelity for unauthorized listeners. On the other hand, in the case of a cryptophony for corded telephone or radio-telephone, if we want to prohibit comprehension, even partial and that we possibly allow ourselves to limit the bandwidth to for example 4 KHz, we can operate with N = 64, Dmin = 50, Dmax = 3200, T = 0.1 s, which makes the channel very protected. Obviously, these parameters being instantly programmable, we can consider defining variable encryption styles that we can apply dynamically during transmission. This possibility allows operation to be less hardened in the case of a transmission in very disturbed environments. If the quality of the transmission improves, the level of hardening of the encryption may increase. We can also discourage the construction of pirate decoders by preparing different sets of parameters which will be applied, for example, in more complex ways each semester.

Among the difficulties linked to the principle of the invention, we have two problems which we have to solve:
1- synchronize the moments of change in modulation period between the transmitter and the receiver 2-transmit or not the starting keys in the pseudo-random registers providing the bits representative of the encryption levels.

In the case of bidirectional cryptophony such as the radio telephone for example, a transmission sequence can be established at the connection to transmit the keys or its encrypted representation. This sequence, carried out in clear, will continue by the synchronization of the two generators, and only the moments of frequency change can possibly be cropped in time by the transmission of a clock in the form of a particular modulation. Optionally, this exchange of keys can be done at a different time from the conversation exchange so as not to sign the encrypted message. It is also possible not to transmit the keys if the two transmission systems have previously agreed on a key depending for example on the day, the hour or the correspondent's number. As we can transmit by a sub- carrier, as described below according to the invention, the information of instants of change, we only have to guarantee a sequence synchronization for example to more or less a second. The reception system receiving, for example, a clock top every second, it will suffice to guarantee a synchronization of the changes to more or less 500 microseconds.

According to the circuit configurations, our experiences have shown that the system could in any case tolerate without noticeable distortion, a delay in the change of about ten modulation periods. We can very easily guarantee a frequency drift of the radio telephone clock of one second per day if we do not want absolute time setting at the start of the conversation.

 In the case of using the method of the invention by a pay TV channel, we have several possibilities to synchronize these moments of change.

For example, if we modulate at 12800 Hz, a frequency chosen by the Pay TV channel Canal Plus, the period is 78 microseconds, the tolerable delay is 780 microseconds. The presence of the television signal gives us a time synchronization framework which is precise to the nearest line, i.e. 64 microseconds for the French standard. We will therefore have no problem in making, for example a change every 50 frames, if we can identify with at least one bit (binary piece of information) the start of the sequence of 50 frames. Currently, the Canal Plus toll chain synchronizes the decoders by a coded line transmitted to each frame, an apparent speed of 50 bauds.

In fact, this line presents a redundant sequence which not only contains a single sequence of synchronization of packets of frames, but it also transmits encryption levels. It is very easy for a person skilled in the art to use this slightly transformed sequence to transmit not only the synchronization of packets of frames which will serve for the more general framework of synchronization of the changes in modulation period, but also1 to transmit an image. coded addresses and starting words of pseudo-random registers which will provide varying levels of encryption over time, This is a problem specific to unidirectional cryptophonic transmissions with a single transmitter and a plurality of receivers which, randomly switching on, must synchronize the flight with the transmitter.

Obviously, the type of transmission according to the invention is not specifically dependent on this or that television standard (Pal,
Sécam, NTSC, Mac or D2Mac), but is also not dependent on the signal transmission means: cable, terrestrial hertzian broadcasting or sati'rite because this process accommodates any standard sound channel possibly with a pre-emphasis acute frequencies on transmission.

 Another advantageous embodiment of the invention is to link dynamic cryptophony to dynamic cryptography. That is to say that if we consider that for a toll chain, all the effort of encryption and transmission and synchronization of the procedures is done for the image according to multiple methods described or used, it is simple to '' use a very small part of the bits (binary information elements) generated, transmitted, received and synchronized for image cryptography. In this case, it is sufficient to agree on the design of the system, that for example, every 50 frames, the first 4 bits representative of the image encryption will be used for the sound encryption on 16 levels. In this case too1 the quality of the secrecy of the cryptophony is completely dependent on the quality of the secrecy of the image. A pirate will have to decode the image to decode the sound. Another means of transmitting the cryptophony bits is to use a digital transmission channel according to already existing standards, for example transmission. of the DIDON type which occupies a few lines at the start of the image and which is at high bit rate, or else a type of duobinary encryption such as those used in European television standards of the MAQ type.

 With regard to the synchronization of the period change times, it is preferable to transmit in advance the next values corresponding to the next periods to be applied. In fact, we can prepare from one second to the next, in, for example, the decryption management microprocessor, the calculations to obtain the next period value; rather than operating an abrupt order which might perhaps be badly transmitted, it is possible to prepare the work to be done and then execute it at a tempo. In another concern to check the good functioning of the system, it is possible to envisage to continuously transmit step-by-step variations of the word representative of the level of encryption in order to allow, at the same time, less sudden jumps in frequency and, a verification over several periods of the fact that the preparations of values have indeed been transmitted. If a period value planned for a future transition deviated by more than one step, it would be possible to replace the defective step (badly transmitted) by an interpolation between two steps located on either side of the defective step. This system would also be usable in the event of a momentary interruption of the transmission by radio-telephony, because each coding and decoding process would live its own life without effective synchronization. If we reestablish the communication on the receiving side, a recovery process by interpolation will not lose synchronism. If, we could not recover lost bits, the communication would only be scrambled for a short time.

 In the case where the object of the invention applies to the cryptophony of a transmission channel which is not connected to a signal on which we can easily synchronize like television, it is possible to use the fact that the audio signal is strictly limited in frequency on transmission as on reception. In a provision of the invention, we transmit the binary and synchronization information via a carrier located outside the useful frequencies, for example above the modulation frequency, but below the maximum audio frequency allowed on the transmission channel.

 Those skilled in the art know several means of transmitting a bit rate of at least 10 baud, more favorably 100 baud and of mixing on this so-called signaling channel, the clock, a submultiple or an overmultiple of l 'period change clock, and the binary data relating to the encryption principle: starting words of pseudo-random registers themselves coded, nature of these registers,' more or less hardened encryption styles, sets of transmission parameters, absolute clocks or relative. The axis means modulating the signal of the synchronization channel are known to those skilled in the art, we will mention for the record, amplitude modulation, frequency modulation with a limit excursion or phase modulation. The modulation technique can either be by type non-return to zero or any other known method or be represented by more than two levels such as duo-binary modulation. The means of receiving this modulation are known: for amplitude modulation, either a level detector preceded by a rectification and an integration or else a peak detection or an all-or-nothing carrier detection. Or, for frequency or phase modulation, a phase locked loop system with or without a divider in the loop, or a phase shifter with its associated detector.

 Another means made to manage the transmission by carrier is to use a programmable counter possibly included in a monolithic microprocessor (Monochip in English). In this case, we can advantageously use the transmission, the frequency generation function to provide the carrier. This generation of frequency can obviously be variable and wet the carrier either in phase, or in frequency, or by an all or nothing control bit. This carrier is mixed with the dynamically returned audio signal. On reception, an additional high-pass filter isolates this carrier and after detection of zero crossing, or high gain amplification, it is presented at the input of the programmable reception counter which is located in the monolithic reception microprocessor. programmable counter is programmed in frequency counter mode and will thus detect either phase variations or frequency variations. In the case of amplitude modulation, a tuned bandpass filter can be used with a conventional detector which will be connected to one of the bits of an input port of the microprocessor or to the interrupt line.

In another configuration of the invention, in order to simplify the task of the monochip microprocessor and the software, it is envisaged to use a programmable asynchronous communication interface (Asynchronous
Communication Interface Adapter in English or CFIA). This type of interface, as well as programmable counters are very often present in monolithic 8-bit microprocessors. The use of these circuits does not therefore entail any additional costs once we have installed the microprocessor which is in any case highly recommended for reasons of space, simplicity of implementation, cost and dynamic programmability. in conjunction or not, with the programmable counter can on transmission, encode the carrier according to 8-bit words framed by a start bit and a stop bit according to a NRZ coding type and at a transmission speed of, by example, 150 baud.

On reception, a similar system will process the asynchronous signal after demodulation. It is finally possible to work with microprocessors which have in situ not only an ACIA, but a real integrated MODEM. The only constraint is to allow them to work at a modulation frequency of the order of 15 kHz for example, located in any case above the modulation frequency.

 In an effective use of the invention in radiotelephony, in radio or pay-TV in which the transmitter would decide to obtain a maximum of guarantees on the identity of the correspondent, a great security and a possible modification very quickly of the encryption algorithms pseudo-random, it is envisaged to use a chip or memory card which would contain all the information intended for the eventual identification and the access titles to the transmission system. If the bit sequence generation system pseudo-randomly came to be known to the public, it is very easily possible to send authorized users a new card containing a new algorithm without changing in any way the decoding equipment and the circuits according to the principle of the invention.

Obviously, the designer of the new smart card would know the technical limits of the method according to the invention, parameters acceptable by the transmission network and would consequently adapt the dynamic performance of changes in encryption levels.

 On the other hand, the circuits according to the invention can be used according to several configurations: in radio-telephony for example, it is possible to switch the coding system to the decoding system, as we will see in the description. This saves on circuits in the case of a simplex transmission. It is also possible to transmit on a clear channel and receive on the other in encrypted or vice versa to have only a half-cycle encryption circuit. In full-duplex transmission, two circuits will be used, but it is possible to have the same synchronization and the same synchronous encryption algorithm for the two simultaneous processes.

 One of the applications of the method according to the invention is cryptophony applied to a transmission on an audio channel but which does not carry speech information: for example and not limited to these, voice frequency dialing (Dual-Tone Multi- Frequency, DTMF in English), transmission by MODEM (Modulator-Demodulator) between two computers or terminals, facsimile transmissions on the switched telephone network or radio, radio teletype, etc. In these types of transmission, the fact of shifting the spectrum dynamically leads to an impossibility either to recognize the precise signaling frequencies, or to follow the modulations-demodulations of data. The fact that these systems generally operate according to very strict standards of frequency precision, makes it possible to carry out a light encryption , with Dmax = 500 for example, to completely jam an unauthorized reception system.

 It is also possible with this system to prevent a user from entering any computer or telephony system since signaling becomes variable in frequency, the hacker will not even be able to connect to the system. The decisive advantage of the process according to the invention is that it respects the standards, the existing installations by its fidelity of restitution and that it can be installed afterwards if we want to harden an existing installation.

 Another application according to the invention is to be able to record the encrypted signal on any medium known to those skilled in the art, mainly magnetic, optical or on a computer memory, according to analog or digital methods. This encryption and its digital representation on the signaling track can be stored together or separately on another analog track or on a complementary digital track such as, for example, the sub-codes of a compactdisc. In this way, it is possible to carry out an encrypted storage of audio frequency, speech, music or data information, to provide this information to a plurality of users without risking that it will be illegally decoded or to allow everyone to has access only to information that concerns him.

 The description which follows, referring to the appended drawings describing nonlimiting examples will make it clear how the invention can be implemented.

 Figure 1 is the overall block diagram of coded transmission.

 FIG. 2 describes the principle of the single fixed reversal decoder of the Canal Plus type in its first embodiment.

 FIG. 3 is the set of spectra of the signals processed according to the fixed frequency inversion methods.

FIG. 4 is the set of spectra of the signals processed according to the method of the invention at variable frequency
FIG. 5 describes the principle of a decoder with variable return frequencies using filters with switched capacities.

 Figure 6 describes the precautions for using an FCC.

 Figure 7 is the diagram of the modulator by digital switches.

 Figure 8 is the exploded diagram of the variable frequency generator used according to the principle of the invention.

 FIG. 9 is the diagram of a programmable commercial counter usable for the principle of the invention.

 FIG. 10 is the complete block diagram of the transmission system with variable turning frequency according to the subject of the invention.

FIG. 11 is the diagram of an encoder or a decoder according to the principle of the invention
FIG. 12 is the diagram of an encoder or a decoder according to the principle of the invention, all the filters of which are of variable frequency.

 FIG. 13 is a diagram according to the principle of the invention where we transmit on a signaling carrier, the instants of change of encryption level and possibly the digital data relating to the pseudo-random generations.

 FIG. 14 is a diagram of the use of the invention where all the useful frequencies are generated by the same oscillator controlled by a quartz resonator.

 The device of FIG. 1 recalls the overall principle of an encrypted transmission by spectrum inversion, the audio input 1 is connected to a low-pass filter 2 which is connected to the modulator 3 whose input 5 presents the modulation frequency ; the output of the modulator 3 is connected to a second low-pass filter 4 intended to eliminate the upper modulation band; the filter 4 is connected to a radio frequency transmitter 6 which passes through an encrypted radio communication 7 which is received by a receiver 8 whose audio frequency output is connected to a third low-pass filter 9 connected to a modulator 10 whose input of modulation 12 has a frequency similar to that present in 5; this modulator 10 is connected to a fourth low-pass filter 11 which supplies the signal 13 with a clear signal.

In this classic configuration, the different low-pass filters have different roles: the first filter 2, filters the audio band to avoid an overlapping distortion between the acute frequencies present in the signal t and the modulation frequency 5; the filter 4 is intended to eliminate the upper modulation band from the modulator 3; its cutoff frequency must be centered at least on the modulation frequency 5, and must have a very steep cutoff to avoid transmitting frequencies beyond the modulation frequency 5 which, if they were present on the modulator 10, would also lead to overlapping distortion; the filter 9 must symmetrically to the role of the filter 2, eliminate the parasitic frequencies which could have been added by the process of radio frequency modulation 6 and 8, or parasites on the transmission 7, or crosstalk residues between the cables located between 8 and 9 in the case of encrypted transmission on TV media; the filter eliminates the upper modulation band and must in any case eliminate the carrier 12 if it is in the audible spectrum. Adjustments to these general rules can be favorably made if we are for example in the case of unidirectional cryptophony because in this case the cost of the encryption device from 1 to 4 is unique and can be very efficient, while we will seek to reduce the cost of decoders 8 to 13 which will be very numerous.

Figure 2 shows the block diagram of the decoder used by the chain
Canal Plus for around three million set-top boxes manufactured to date. The decryption circuit between the audio input 14 and the output 13 is formed by a first low pass filter 9, a modulator 10 and a second low pass filter 11. The modulator 10 is controlled by a fixed frequency 12 of 12800 Hz reference obtained by a phase-locked loop constituted by a block 16 and a digital frequency divider 17. Block 16 consists of a phase detector followed by a low-pass filter itself controlling an oscillator This phase locked loop synchronizes in phase on input 15 which is a frequency at 50 Hz from the television frame frequency; The slave oscillator therefore oscillates at 50x256 = 12800 Hz, which is the inversion frequency.

 Figure 3 explains the spectra obtained at each stage of signal transformation. We have chosen as an example the inversion frequency of 12800Hz, but this principle applies whatever the fixed frequency chosen. The spectrum A represents the output of the first low-pass filter when we want a frequency reversal at 12800Hz.

The filtering should theoretically be flat up to the maximum frequency of the passband chosen by the transmitter, for example 10500Hz for a television transmission, and cut, at, for example, -60 dB the frequencies from the frequency of modulation. The B spectrum represents the result at the output of the modulator: band inversion around the modulation frequency. The spectrum C indicates the spectrum after filtering of the second low-pass filter which will theoretically between flat until the modulation frequency and cut all the frequencies beyond. This theoretical filtering being by definition impossible to carry out, we will tolerate for example, a practical cut-off frequency of 11000Hz and will use a filter of the 12th order, if we are in the case of a high quality diffusion.

The advantage of filtering very vigorously in C, is to simplify the subsequent filtering by the third low-pass filter which can cut from
13000Hz and be only of the seventh order. The D spectrum represents the result at the output of the second modulator: band inversion around the modulation frequency. The E spectrum is the final spectrum from the fourth low-pass filter which is at least of the 7th order and preferably of the 9th order where we mainly seek to eliminate the carrier frequency which is a very unpleasant residue to hear.

 Figure 4 represents the spectra obtained if we decide to vary the modulation frequency by way of example from 12800Hz as in the previous example to 13800Hz. The spectrum F represents the output of the first low-pass filter which will be equivalent to the spectrum A. The spectrum G shows that, at the output of the modulator, a variable frequency of modulation upwards translates in block the spectra upwards. The spectrum H is that which we will have to obtain after the second low-pass filter if the modulation frequency is positioned at 12800Hz, as in the spectrum C. If the modulation frequency is at 13800Hz, we will have to have a curve equivalent to that presented by the dotted spectrum 1. A question arises, if we filter too abruptly on the second low-pass filter, the signal will lose acute frequencies at this stage; but these treble frequencies from the first modulator are representative of the bass frequencies after the second reversal. Therefore, filtering the high frequencies from the first modulator too much amounts to filtering the low frequencies perceived by the listener too much. We therefore have several possible choices: limit the low spectrum by filtering with a high-pass filter before the first low-pass filter, which is only possible for medium quality transmission, or limit the frequency excursion Dmax of the carrier while remaining in a reasonable zone with filters with fixed frequencies, or else, varying the cut-off frequency of the second low-pass filter to follow the changes in the modulation frequency. In this case, we have the variable spectrum I to the outcome of the second low pass filter. The J spectrum shows the spectrum at the end of the second modulator and the K spectrum is the spectrum resulting at the end of the fourth low-pass filter. It should be noted that the third low-pass filter can advantageously be fixed if it passes, without excessively attenuating, the modulation frequency; if it attenuates the signal too much by the fact that its cut-off frequency is for example fixed at 12800Hz, the resulting effect on the signal perceived by the listener would be a variable suppression of the low frequencies according to the modulation frequency, as a high pass which we would continually vary the cutoff frequency in the bass .. Naturally, = according to the principle of the invention, if the third low pass filter is variable with the modulation frequency, the transmission will be that better. In a certain embodiment according to the invention, we can therefore perfectly tolerate the fact that the third low-pass filter is at fixed frequency, to simplify the circuits and reduce the costs, this without causing distortion on the signal on the condition that the frequency excursion is such that the maximum modulation frequency is inside the band passed by the filter, but nevertheless that the filter attenuates the parasitic frequencies sufficiently, in particular in the case of a TV transmission, the line frequency of 15625Hz.

 FIG. 5 represents a new variable modulation frequency decoder using switched capacity filters (FCC). The decryption circuit located between the input 14 and the output 13 consists of an FCC filter 9, a modulator 10 and an FCC filter 11; the modulation frequency of the modulator 10 is obtained by a selector 22 controlled by a control signal 23 which positions either a fixed frequency coming from a divider 21, or a variable frequency present on an input 24; The reference frequency generator 18, preferably quartz, has an oscillation frequency divided successively by two with the divider 19 which feeds the FCC filter 9, then by another divider by two 20, which feeds the FCC filter 11.In a possible configuration of the decoder, we can have an oscillator 18 at 3.276.800Hz which provides the reference frequency from 9 to 1638400Hz to obtain a cut-off frequency of 1638400/100 = 16384Hz because in a possible realization of the FCC, it is advisable to provide at the low-pass circuit, a reference frequency one hundred times greater than the cut-off frequency of the filter; the cutoff frequency of the second filter will be 8192Hz and the division by 64 of the divider 21 will provide 12800Hz to the modulator. In the following figures, an exhaustive description of the system using FCC filters with variable cut-off sequence will be presented.

 FIG. 6 shows a preferred embodiment of the filters with switched capacities. Between input 25 of the filter and its output 27, we insert on each side of the FCC, an analog low-pass anti-aliasing filter. As is known to those skilled in the art, the digital input of frequency 26 must be, depending on the internal construction of the filter, a hundred times the cut-off frequency of the low-pass filter. Since the FCC operates on the principle of sampling, it is essential to have at the input of the filter a signal free of any frequency approaching this switching frequency 26 otherwise, a distortion due to aliasing will be generated, and at the output, it is necessary to filter the signal with an integration function (low pass) to restore the signal without its multiple components of the sampling frequency. We favorably use for this purpose second-order filters composed of on the one hand from R1, R2, C1, C2 and on the other hand from R3, R4, C3, C4. The cutoff frequency of these filters being for example chosen 40% above the maximum frequency to be transmitted. by the filter, we are assured that a dispersion on the absolute values of the components of 40%, will not influence the resulting bandwidth of the FCC plus the anti-aliasing filters. With this precaution, these analog filters are easily integrated into a monolithic circuit, as proposed by some silicon smelters on demand, known to those skilled in the art.

 FIG. 7 represents an embodiment of a digital modulator as used in one of the embodiments of the invention. Between its input 28 and its output 29, there are two sets of analog switches controlled by femieture signals 30 and 31 which are in phase opposition. This circuit is in two states; One being: 30 high and 31 low, The other being 30 low and 31 high.

The general gain of the circuit is -1 in one state and +1 in the other. One of the characteristics of this circuit is that the accuracy of this gain is dependent firstly on the ratio of the two resistors connected to the negative pole of the operational amplifier and on the ratio of the two resistors connected to the positive pole of the operational amplifier and secondly on the ratio between the resistance in the on state of the analog switch with respect to the value of the resistance R. If we wish to integrate this modulator in a monolithic circuit, it is easy to obtain ratios of two resistances guaranteed at 1%, even if the absolute value of the resistance R is more or less 40%. On the other hand if the resistance in the on state of the switch is guaranteed less than or equal to 100 Ohm and that R is for example 1 Megohm, we can so overall predict a differential gain error of less than 1%. In this case, this modulator not only has the decisive advantage of being easy to integrate into a monolithic circuit on demand, but also, it presents a rejection of the carrier at guaranteed output better than 40 dB, which leads to not not make additional manual or automatic adjustments to improve this rejection. Those skilled in the art know that in the other types of circuits used to produce a modulator, there must be at least one adjustment intended to improve the rejection of the carrier. This is possible, either by adjusting the medium point of action. of the differential current of an analog modulator by current multiplier, that is to say in many other cases, a part of the carrier is reinjected in phase opposition with this one at modulation output, and we are obliged to adjust the amplitude or the phase of this reinjection. On the other hand, when we use a classic analog modulator, we have to provide the modulation input with a sinusoidal signal, so we have to filter the generally square signal coming from the oscillators and the dividers. In the case where we use a modulator according to FIG. 7, a favorable embodiment of the invention, we control the inputs 30 and 31 by a direct digital signal without filtering. Naturally, we can use to achieve the object of the invention, a modulator different from this and known to those skilled in the art, but with the difficulties described above.

FIG. 8 describes the principle of the variable frequency generator used for the object of the invention. The variable frequency present at output 43 is obtained by counting down a master clock 40, preferably obtained directly or indirectly from an oscillator regulated by quartz; This down-counter is made up of two eight-bit down-counters 37 and 38 connected by means 39 in series in such a way that they form a sixteen-bit down-counter and whose output 41 is looped back so that the values stored in the registers 32 and 35 are reloaded into the downcounters via their parallel inputs when the downcount is zeroed. In other words, the downcounting cycle begins with a parallel loading in the downcounters of the value to be counted down, then each cycle of the clock 40, counts down the value of the down-counter until it reaches zero, signaled by the output 41 which will reset the following cycle indefinitely. A flip-flop D 42 makes it possible to obtain a duty signal 5050 of frequency half. To change the value loaded in the down counter, we must write by the command of loading 34 a value sixteen bits parallel thanks to the inputs of registers 33 and 36. Thus, we have a programmable division circuit of a clock 40 which provides a signal square 43 and whose period can only change at the end of a previous half-period, so that the transitions between any periods generated by a change in the division ratio 33 and 36 will be without periods random intermediaries or phase jumps. This is obtained with an instant of loading by 34 which is asynchronous with the clock 40 or the result 43 because the buffer register 32.35 makes it possible to wait for the right moment to change the period during the signal on 41. We have represented in this figure a down counting register of 16 bits, we could have taken any number of counting bits without changing the object of the circuit; the precision of the frequency hopping, or rather the difference between two generated frequencies depends on the number that we can load in the register. We note that to obtain an output frequency given on 43, the larger the number loaded at 33.36, the greater the division ratio will be and the greater the input frequency 40 will have to be, but the greater will be the precision of generation of frequency. In the case which concerns us, that is to say variable frequency cryptography, the difference between the different frequencies generated does not have to be according to such or such value, it suffices that the coder and the decoder act on the same frequencies known in advance. On the other hand, to avoid any distortion, it is essential in the subject of the invention that these frequencies are obtained precisely and instantaneously and without errors in the transitions between two frequencies, which is the case with this generator, unlike the prior art which limited one or more of these parameters. Without changing the object of this circuit, configurations dependent on it can be obtained with a number of bits different from the buffer registers and the down-counter; we can optionally load only the register 35 containing the bits of low bit weight and leave the register 32 fixed if it does not have to change at each loading; we can possibly eliminate the D 42 flip-flop, if we do not wish to provide the modulator with a square signal; we can possibly use a counter instead of a down counter and it would generate an output signal when the counter which starts from zero reaches the maximum value stored in the buffer register; we can finally use any programmable counting means with a digital value
N which would provide a signal such that the output frequency is equal to the input frequency divided by a divider N and having at least one means of charging N in such a way that an abrupt change in N causes a change in period at from the current period.

 FIG. 9 presents a means of frequency generation by programmable generator usable for the object of the invention as described by the manufacturers of integrated circuits peripheral to microprocessor systems and known to those skilled in the art. A sixteen-bit counter 49 will count, in a specific operating mode of the circuit, the pulses present on the input 50 and will provide, thanks to the control logic 47, a square signal on the output 51. The control logic 47 contains no only the output flip-flop described above but can also, under the cover of the control register 48, generate a sequential loading pulse of the eight-bit registers 44 and 46 which are connected to the bus of the microprocessor 45. ia procedure for loading the registers is done twice because in most cases the bus is 8 bit wide. The control register and the control logic will make it possible to recognize the address of the register concerned by the writing memory of the instruction of the microprocessor and to load in the right way the most significant and least significant byte. In some programmable counting circuits (Programmable Timer in English), double buffering is provided upstream of the counter to load registers 44 and 46 at once with sixteen bits at once because otherwise, if we are in a configuration of sequence where we have loaded only the most significant bit byte and the least significant bit byte is not yet modified at the time when the sixteen bit counter takes its down count value, we will have an incorrect value entered.

 Figure îC: shows the complete variable frequency cryptophony system. The encoder between its input 1 and its output 58 comprises a low-pass filter 2 with fixed frequency 52, a modulator 3, a second low-pass filter with variable frequency 53 generated by a phase-locked loop 55 comprising a divider in the loop whose input is dependent on the modulation frequency 5. This frequency 5 is supplied by a programmable generator 54 which divides a reference frequency 56 by a variable division ratio N present on its input 57. The output of the encoder 58 is connected to a decoder by its input 61. We obviously imagine that between the output of the encoder 58 and the input of the decoder 61, we have transmitted (or stored and re-read) the encrypted signal. The decoder located between the input 61 and the output 13 includes a low-pass filter 9 with variable frequency 59 , a modulator 10 and a second low-pass filter f1 at fixed frequency 60.

The variable frequency 59 is obtained by a phase locked loop 65, comprising an internal divider, controlled by the modulation frequency 12. This variable modulation frequency 12 is obtained by a programmable divider composed of a counter 62 which divides a frequency reference 63 according to a division ratio N supplied by the input 64. As described above, the encoder between 1 and 58 is located at the transmitter and finally comprises as variable parameter only the number N present on the input 57, all frequencies being generated from this value. The decoder between 61 and 13, is located at the signal receiver and has as variable parameter only the number N present on the input 64, all the frequencies being generated from this value. To fix the ideas and as non-limiting example of the operation of this circuit, we indicate below, simplified means for obtaining an average quality of coding and decoding. The frequency of the FCC 52, which we can possibly set for the low pass 2, maybe 819200Hz or a filter of the 7th order cutting the band at 8192Hz and rejecting the frequencies beyond 10000Hz; the clock 56, coming from a quartz oscillator can be 4019200tir in one of the solutions which can generate the frequency of reversal of 12800Hz, already used by Canal Plus; the number N present in 52 can take the values 314 to 34O in such a way that the reversal frequency 5 can vary between approximately 12800Hz and 1t800Hz if we want to modulate in a variable way towards the low frequencies, and N can take the values 314 to 290 in such a way that the inversion frequency 5 can vary between approximately 12800Hz and 13859Hz, if we want to modulate in a va able way towards the high frequencies; Let us not forget, for computation, that if we want a square signal in input of modulator, it is necessary to chain counts of half-periods, and that traditionally in the circuits used the division ratio is of (2Nt2), therefore if we want to divide by 314, the number put in the counter will be 156. The PLL 56 can have in its control loop a divider by 100 so that the frequency generated in 53 is 100 times higher than that present in 5, this which would give a practical cutoff frequency of the FCC equal to that present in 5 in the case where the FCC uses a clock division ratio of 100 times and is of the 7th order; in the decoder, the frequency generation circuit can advantageously be a programmable oscillator (Programmable Timer in English) by the monochip microprocessor which will provide the levels N of encryption on the input 64. The PLL 65 will symmetrically multiply the turning frequency 12 by 100 to control the FCC filter 9 of the 7th order, by a frequency on its input 59, a hundred times higher than its cut-off frequency. The last filter 60 can advantageously be a FCC of the 7th or 9th order controlled by the same frequency as 52 ie 819200Hz. The circuits 62, 63, 64 can be as we have seen integrated into a monochip microprocessor or form an integral part of an already existing circuit as standard in the monochip configuration proposed by the manufacturers of microprocessors. This non-limiting proposition is an example of an inexpensive embodiment with average performances intended for example for the radio telephone. If we wish very good performances for the coder in the case of a circuit at the head of the television network, it is advisable possibly having much higher order filters for filters 2 and 4, especially for circuit 4 which should have an order 12 with variable frequency, it is obviously a delicate circuit to realize but which is only present 'Once. On the other hand, if we wish an economic realization for the multiple decoders, a 7th order for the filter 9 and a 9th order for the filter 11 are sufficient. We can also tolerate for the purpose of simplification and economy, with all the precautions mentioned before, to leave the frequency of the filter 9 fixed if we modulate according to the restrictions described, which removes on each decoder the phase locked loop which is difficult to integrate at very low cost in the current state of the art.

 FIG. 11 represents a decoder according to the subject of the invention with variable reversal frequency usable for the encryption of a television sound channel. Between input 58 and output 13, there is a low frequency filter 9 with variable frequency 59, a modulator and a low frequency filter with fixed frequency 11, controlled by 60. The programmable frequency generation circuit 68 includes a programmable divider 62 which counts down a clock 63 according to a value 64 which is presented to it. This value 64 is obtained by a pseudo-random sequence generator circuit 67 which is supplied by digital control words 66 which come from digital transmission on the television medium. This medium provides the frame and line synchronizations to supply to the circuit 62 the division values 64 at the right times synchronized with the circuit present in the encoder. On the other hand, circuit 67, which can advantageously be a programmable microprocessor, recognizes the encrypted information supplied by 66 on the TV channel to decide which pseudo-random sequence to use and where to position itself in the pseudo sequences.

 FIG. 13 presents a block diagram of an encrypted transmission where we indicate the means for achieving synchronization between the transmitter and the receiver of the instants of change of frequency of reversal by the transmission of a signaling channel. The encoder has the circuits already described 1,2,3,4, it further includes an audio mixer 73 which adds to the reverse encrypted signal, a signaling carrier generated by an oscillator 77. This oscillator can be modulated by a command from the circuit the encryption master may be another divider programmable by a division ratio.M of the same clock 75 as that supplied to the programmable division circuit 76 for the generation of reversal frequency. The transmission 6 and reception 8 means convey the encrypted signal and the signaling carrier by a radio channel 7. the decoder has the circuits already described 9,10,11 and more generally has a means 80 already described for controlling the frequencies of the filters 9 and 11 whether or not from a turning frequency 12 obtained with a programmable clock 82. But this decoder also has a circuit for receiving the signaling carrier composed of a high pass filter 83, a means of reception of the modulation of the carrier 84 and of an amplitude, frequency or phase modulation detector 85 which can supply the value M to the decryption control circuit which can favorably be a programmed microprocessor circuit. and by way of nonlimiting example, we have a programmable divider in circuit 77 which generates two different frequencies according to two values of M, the square signal resulting is filtered and mixed in mixer 73 to provide a frequency modulated sinusJidal carrier; the receiving circuit 84 amplifies and detects the zero crossing of the carrier signal and provides a digital signal to the circuit 85 which is a programmable counter used as a frequency meter, so that it provides, at each period of the signaling carrier, a indication of its frequency and that a process thus detects these transition moments. In another optional embodiment shown in the same diagram, the signaling channel is modulated by a signal from an ACIA 78, asynchronous serial communication interface, generally present in a single chip microprocessor circuit and which modulates according to the same means 79 as those described above, the signaling carrier, so that, in the decoder, if we choose a frequency modulation for the signaling carrier, we can use a phase locked loop 87, provided with a divider 86 which will supply the ACIA circuit 88, the data he digital asynchronous series which will be recognized. All these block diagrams contain circuits known to those skilled in the art, this is why we will not go into detail any further, the principle of an asynchronous serial transmission by modulator-demodulator being well documented in the technical literature.

FIG. 14 represents a proposal for a very simplified decoder capable of receiving the signals transmitted according to the principles of the invention. The decoder circuit, preferably integrated into a specific integrated circuit, comprises between its audio input 58 and its audio output 13, a first low-pass filter FCC 9 controlled by an externally generated frequency 59, a digital modulator 10 according to the diagram in FIG. 7, the frequency input 12 comes from a programmable divider 93, a second low-pass filter FCC 1 1 controlled by the frequency 60 coming from a divider 95 controlled by the input 59. To control the different frequencies necessary for the organs decoding, we can use a generator 89 controlled by a quartz 90 which produces the frequency F0, then by division by 5, the frequency F3 intended for the programmable divider 92, and by division by 16, the frequency
F1 intended for the decoding circuit 96. The circuit 92 can advantageously be a masked monochip microprocessor 93 comprising a programmable divider 94 which divides F3 to give variable FX as a function of time. To fix the ideas, by way of nonlimiting example, we we can imagine that FO is at 26.625 MHz, frequency chosen to allow analog-digital conversion to video frequency on the encrypted signal which will allow, thanks to a fast digital processor, to operate a decoding algorithm and to render by a digital-analog converter, the decrypted signal. Fl will therefore be at 1.664 MHz and F2 at 832 KHz. F3 will be at 5.325
Mhz and by a division by 416 could give FX = 12800 Hz for possible reasons of compatibility with existing systems. We will be able by varying this programmable divider, to obtain frequencies
Variable FX within the limits described above. Variants of this circuit make it possible to integrate not, for example, the divider 91 in an ASIC 89 but rather in circuit 96, the distribution of all these pieces of circuit being able to vary according to the integration possibilities that we have available. On the other hand, the fixed division ratios of 89, 91 and 96 are given as an indication. Describe here all the combinations of division ratio which would make it possible to obtain on the one hand the driving frequencies of the FCC filters and on the other apart from a fixed reference frequency F3 and programmable division ratios by 92 and 94 is impossible
During the practical realization of the circuits of the different figures, we have successfully used:
In all the figures, the low-pass filters with switched capacities can be produced with TSG 8512 from Thomson (currently called ST) for filters of the seventh order and TSG 8514 for filters of the 8th order or combinations of These, Thomson offers, among others, to carry out FCCs according to oversized templates up to 12 th order. It is also possible to use circuits
National Semiconductor MF10 (5th order) or MF6 (6th order) for
FCC. The switches are for example CMOS type circuits
Motorola MC 14066, or MC 14053, the phase-locked loop oscillators can be of the Motorola CMOS type MC14046, the frequency dividers and CMOS counters: 4013, 4018, 4520, 4060, 4522, 4024 depending on the division ratio. The registers can be CMOS MC 14013, MC 14174, MC 14034. The crystal oscillators can be 4060 or using 74HC04.

Operational amplifiers are either integrated into FCCs, or of the type
TL084 with high input impedance. For the embodiment of the modulator of FIG. 7, the resistors R can validly belong to an integrated network of resistors, which makes it possible to obtain at low cost a set of resistors whose differential dispersion is low.

The integrated frequency dividers (programmable timers) in particular in FIG. 9 connectable to the microprocessors are of the Motorola MC 68B40 or Intel 8253 or preferably 8254 type. The ACIA circuit of FIG. 13 can be the Motorola MC 68B50, or Zilog 8530. The microprocessors successfully used are of the Intel 8048 or 8051 or 8096 type or in the Motorola MC 6805 family, or Hitachi Ho64180; a microprocessor having inside the same case, both a programmable divider and a serial communication circuit is highly recommended.

Claims (20)

Claims 1. Coding method for secret transmission of  audio signals, consisting of  - perform a first low-pass filtering of the signal  -modulate the signal thus filtered using a  variable frequency generator  - perform a second low-pass filtering of the  signal thus modulated, characterized in that the signal is modulated using a variable modulation frequency in discrete steps. 2. Device according to claim 1, characterized in that the modulation frequency changes are made at the end of an entire period of the modulating signal. 3. Device according to claim 1 and 2, characterized in that the frequency generator uses a parallel loading counter which is loaded with a count value at each half-period of the modulation signal. 4. Device according to claim 1 to 3, characterized in that the frequency generator is a programmable frequency divider which comprises a counter, means for loading this counter by a register itself comprising loading means, which causes the generator to change period at the end of its previous period regardless of when the register is loaded.  5. Device according to any one of claims 1 to 4, characterized in that the second low-pass filter is a filter with variable cutoff frequency depending on the modulation frequency. 6. Device according to claim 5, characterized in that the filter with variable cut-off frequency is a filter with switched capacities. 7. Device according to claim 6, characterized in that, the switched capacity filter is controlled by a frequency multiplier composed of a phase locked loop comprising a frequency divider in the loop, multiplying the modulation signal by a fixed value. 8. Device according to claim 6, characterized in that, the switched capacity filter is controlled by an oscillator divided by a programmable divider. 9. A decoding method for the secret transmission of audio signals transmitted by the device according to any one of claims 1 to 8, consisting of  - perform a first low-pass filtering of the signal  -modulate the signal thus filtered using a  variable frequency generator  performing a second low-pass filtering of the signal thus modulated, characterized in that the signal is modulated using a variable modulation frequency in discrete steps. 10. Device according to any one of claims I to 9, characterized in that it is the first low pass filter which is at variable cutoff frequency. 11. Device according to any one of claims 1 to 10, intended for the encryption of the sound channel of a pay-TV transmission, characterized in that the modulation frequency is controlled by a programmable divider whose instantaneous division ratio is obtained by a pseudo-random generator whose starting words are transmitted by a digital channel present on the medium of the video transmission. 12. Device according to claim II, characterized in that the frequency changes of the modulation are synchronized on a frequency submultiple of the frame period of the video transmission medium. 13. Device according to claim 8, characterized in that the pseudo-random generation of modulation frequencies is controlled by part of the bits of a pseudo-random generator emanating from a digital coding device of the video transmission. 14. Television receiver, characterized in that between its audiofrequency detection circuit and its audiofrequency amplifier, a device according to any one of claims 11 to 13 is inserted. 15. Method according to any one of claims 1 to 14 characterized in that the synchronization of the instants of frequency changes between the transmitter and the receiver is ensured by the transmission of a modulated carrier located at a frequency higher than the main modulation frequency which is added to the modulated signal on transmission and filtered by a high-pass filter on reception 16. Device according to claim 15, characterized in that the carrier carrying the synchronization information is frequency modulated and that a phase locked loop will be used for demodulation on reception. 17. Device according to claim 15, characterized in that the carrier carrying the synchronization information is phase modulated and that a phase locked loop will be used for demodulation on reception. 18. Device according to claim 15, characterized in that the carrier carrying the synchronization information is amplitude modulated and that a threshold level detector will be used for demodulation on reception. 19. Method according to any one of claims 15 to 18, characterized in that the synchronization carrier also carries representative bits for the synchronization of the encryption registers at transmission and at reception intended for programming the hops frequency. 20. Radio receiver, characterized in that between its audiofrequency detection circuit and its audiofrequency amplifier, a device according to any one of claims 1 to 19 is inserted.