EP0662268A1

EP0662268A1 - Method and device for automatic gain control in an amplifier of an electroacoustic system, particularly for motor vehicules

Info

Publication number: EP0662268A1
Application number: EP93920901A
Authority: EP
Inventors: Claude Bernard Carpentier
Original assignee: S L X Sarl
Current assignee: S L X Sarl
Priority date: 1992-09-22
Filing date: 1993-09-21
Publication date: 1995-07-12
Also published as: FR2696060A1; WO1994007306A1; FR2696060B1; US5557237A

Abstract

The device comprises both in the processing chain for the complex signal from the microphone (10) and in the processing chain for the effective signal output from the gain-controlled amplifier (1), a first high-pass filtering stage with low switch-off frequency and a second high-pass filtering stage with a medium switch-off frequency and presenting an attenuation of at least 30 dB at approximately 100 Hz. There are further provided first auxiliary comparison means (35) comparing the respective levels of signals taken before and after the first high-pass filtering stage. Said means (35) are used alone in the case where the electroacoustic system is autostable beyond the switch-off frequency of the first high-pass filtering stage to decide on possible blocking of the amplifier gain. Said auxiliary comparison means may be also used in combination with second auxiliary comparison means (38) arranged in the processing chain for the effective signal, or in combination with a equalizer (7).

Description

Method and device for automatically controlling the gain of an amplifier of an electroacoustic system, in particular for a motor vehicle.

The invention relates to the automatic gain regulation of a gain amplifier controlled by an electroacoustic system as a function of the ambient signal / noise ratio.

It finds a particularly advantageous application in the compensation of ambient noise of electroacoustic audio frequency chains with gain-controlled amplifier, in particular with several channels, in particular those incorporated in motor vehicles.

Solutions or attempts at very different solutions have been proposed in this area, to try to solve a very important problem with this type of control device, namely the ^{identification} of the nature of the signal detected by the sensor device. This captured signal can in fact be noise produced by the car, the signal from the loudspeakers or a mixture of the two in a greater or lesser proportion. However, and suπout if this sensor is a microphone, the slightest error in the discrimination between the signal and the noise immediately leads to a positive reaction, the signal self-amplifies on itself and the system quickly reaches its maximum gain in a process comparable to the LARSEN effect. In order to get around this difficulty, it was proposed at an AES congress in 1988 to use as a sensor a vibration detector fixed to the body of the car after having highlighted a certain correlation between the noise inside. of ^the passenger and vibrations of the body. Besides the fact that this solution is not ideal because the correlation is only partial, it does not completely eliminate the problem of signal and noise discrimination because the sound waves coming from the loudspeakers cause vibrations of the body which excite the detector.

^Other devices utilize, in the signal processing chain from the microphone, a low pass filter ^of high order with cutoff frequency of around 15 Hz, which radically eliminates waves from loudspeakers whose frequency is always higher. These devices are not efficient because they _. also eliminate the components of the noise spectrum located in the range of audible frequencies to keep only the inaudible components which are not disturbing. For these systems to function properly, the level of the annoying noise frequencies would have to be proportional to the level of the frequencies below 15 Hz. Which is absolutely not the case. Finally, a third method widely used in public address systems consists of measuring noise in the absence of a signal. These systems work very well when the signal is "chopped" like for example the human channel which leaves frequent periods of silence during which it is possible to carry out without possible error measurement of the noise. But when it comes to music, there can be several minutes without a single moment of silence, which eliminates this type of device for the type of application targeted.

In devices intended for the automobile, the best results are currently obtained in the following manner. A microphone captures a complex signal composed of car noise and sound modulation from the speakers. This signal is filtered through ^a high-pass filter for removing very low noise frequencies and which are not troublesome for the hearing and is then rectified to be compared to the useful signal or pure taken after the ^amplifier gain controlled, filtered by an identical filter and rectified. A calibration comparator pink noise modulation is carried out beforehand in the ^absence of ambient noise, so as to make comparable the two voltages from the respective rectifiers pure signal channels and complex signal. The problem raised by these. devices is the non-linearity of the speaker / room / microphone electroacoustic chain. A response curve record at the microphone shows variations around an average value which commonly reach 15 dB. These 15 dB represent the uncertainty of these devices in distinguishing the signal from the noise. Also, if we don't want them to confuse the signal and the noise, during the adjustment, a safety margin of 15 dB must be adopted, i.e. ensure that the comparator which controls the increase in gain can only switch if the voltage coming from the micro channel and arriving at the comparator is more than 15 dB higher than the voltage from the pure signal channel.

This safety margin implies that the gain only increases if the level of the partial noise sound pressure is almost 15 dB higher than that of the signal. If the signal and noise spectra were located in the same frequencies, this type of system could not work because the "loudness" of the signal would always remain much lower than that of the noise and the signal would remain inaudible. However, the noise spectrum of a car is very rich in low frequencies which are hardly masking for signal frequencies located in the middle of the sound spectrum like human speech, and it is thus possible to clearly distinguish a signal in a noise environment. who is him

15 dB higher provided that the frequencies that make up the noise are sufficiently different from those that make up the signal.

The disadvantage of this type of system is that they use mainly as noise frequencies frequencies which are not very disturbing for hearing and phenomena which the skilled person often designates by the expression "phenomena of overcompensation" ". and which are interpreted as a system malfunction.

Significant improvements can be made by using an equalizer to linearize the speaker / room / microphone electro-acoustic chain, we can then lower the safety margin to 6 dB. This margin makes it possible to work with a more restrictive filter whose cut-off frequency is 250 Hz and having an attenuation of 20 dB at 100 Hz. Which limits the phenomena of overcompensation by lowering the partial acoustic pressure of noise to 5 dB above that of the signal when the comparator switches.

However, these systems are unsuitable for wide dissemination because, in addition to the persistence of an overcompensation phenomenon, they always work at the limit of stability. This stability is even more precarious when it comes to devices which are the most common in the automotive industry. Indeed, in these devices it is usual to take as the pure signal signal for comparison the sum of the voltages of each channel. This method requires the provision of an additional safety margin of 10 dB. Indeed, in an installation with 4 speakers, when the driver only wants to keep the front left channel because the rest of the family is sleeping and he wants to listen to the information, after the operations of the balance and fader buttons, the pure signal voltage is only a quarter of what it was when the two settings were at their midpoint, a drop of 12 dB. On the other hand, if it is desirable, the microphone is located near the driver's head and, as is common, the front left speaker is located at the left end of the dashboard, the signal component pure reaching the microphone has dropped very little (about 2 dB), because it mainly comes from the front left speaker. We therefore obtain an imbalance of 10 dB at the level of the comparator.

Finally, these systems are provided with a threshold detector device whose function is to prohibit an increase in gain when the signal level drops below a certain threshold, - 40 dB, for example. This device, which prevents an increase in gain between two disc tracks, is however insufficient to avoid an untimely increase in gain between two successive words when a speaker speaks hesitantly in any other similar condition when the signal level does not drop below the threshold but that it is not transiently representative of its level to be taken into account.

All these reasons explain why no automatic volume control system has today reached a level of diffusion capable of satisfying the car radio market.

The invention aims to provide a more satisfactory solution to this problem.

An object of ^the invention is to permit automatic regulation of ^the amplifier using a high-pass filtering having a ^curve very restrictive attenuation in the low- frequencies.

Another object of the invention is to provide a multichannel device with unconditional stability and not subject to overcompensation phenomena or untimely increases in gain on "quasi-whites", therefore suitable for wide distribution in the automotive market.

Another object of the invention is to eliminate the instabilities during the balance adjustment manipulations and to fade in multichannel installations. The subject of the invention is therefore a method of automatically controlling the gain of an amplifier of an electro¬ acoustic system, of the type of that mentioned above, that is to say in which a first treatment is carried out, comprising a first filtering sub-processing, on a complex signal taken from the room using a capture means such as a microphone, and containing a mixture of ambient noise and sound modulation coming from the loudspeakers, so as to obtain a first processed signal; in parallel, a second processing is carried out, comprising a second filtering subprocessing, on a useful signal taken at the output of the gain amplifier controlled so as to obtain a second processed signal; the gain of the amplifier is then regulated as a function of a level comparison between the two processed signals; according to a general characteristic of ^the invention. each filter processing sub comprises a first high-pass filtering in low cutoff frequency so that ^a second high-pass filtering to average cut-off frequency and having an attenuation of at least about 30 dB around 100 Hz ; Moreover, it blocks ^the increase in the gain of the amplifier when the level of the signal taken prior to the first high-pass filtering of the first filtering sub-processing is less than the product of the signal level corresponding taken after this first pass filtering high, by a predetermined coefficient greater than 1. or when the level of the signal sampled before the first high-pass filtering of the second filtering sub-processing is greater than the product of the level of the corresponding signal sampled after this first high-pass filtering, by a coefficient predetermined greater than 1.

It is also possible to solve the problem posed, using an automatic gain control method of the type mentioned above, in which according to the invention, in addition to using a first high-pass filtering at low cut-off frequency as well as a second high-pass filtering at medium cut-off frequency and having attenuation at least of the order of 30 dB around 100 Hz, an equalizer is provided which is adjusted so as to prohibit , in the absence of noise, increasing the gain of the amplifier by any pure (or useful) signal of frequency less than or equal to the cutoff frequency of the first high-pass filtering; this makes it possible to remove the blocking of the increase in the gain of the amplifier when the level of the signal sampled before the first high-pass filtering of the second filtering sub-processing is greater than the product of the level of the corresponding signal sampled after this first filtering high pass, by a predetermined coefficient greater than 1, and to keep the blocking of the gain increase only when the level of the signal sampled before the first high pass filtering of the first filter subprocessing is lower than the product of the level of the corresponding signal sampled after this first high-pass filtering, by a predetermined coefficient greater than 1.

It is still possible to solve the problem posed, using an automatic gain control method of the type mentioned above, in which, according to the invention, there is no need for both an equalizer and blocking the increase in amplifier gain when the level of the signal sampled before the first high-pass filtering of the second filtering sub-processing is greater than the product of the corresponding signal level é after this first high-pass filtering, by a predetermined coefficient greater than 1. Indeed, in this case. and in the event that the electro-acoustic system is freestanding below the cutoff frequency of the first high-pass filtering, that is to say in the event that the electroacoustic chain does not have a resonance specific to put the system in instability, it is sufficient, according to the invention, in addition to using this first high-pass filtering at low cut-off frequency as well as a second high-pass filtering at medium cut-off frequency and having attenuation at least of the order of 30 dB around 100 Hz, to block the increase in the gain of the amplifier only when the level of the signal sampled before the first high-pass filtering of the first filtering sub-processing is less than the product of the level of the corresponding signal sampled after this first high-pass filtering, by a predetermined coefficient greater than 1.

According to a preferred embodiment of the method according to the invention, there is provided a calibration phase in which a pink noise modulation is injected at the input of the amplifier, and a level adjustment is made prohibiting the increase in amplifier gain by the first signal processed, taking into account a chosen safety margin, between approximately 1.5 dB and approximately 4 dB, preferably 2 dB.

This calibration phase then makes it possible subsequently, in a normal operating phase, to trigger an increase in the gain of the amplifier only when the level of the first processed signal will be 2 dB higher than that of the second processed signal. To verify ^the self-stability of the electroacoustic system after ^calibration phase noise in pink, is injected as pure modulation signal. a sinusoidal signal of variable frequency, the frequency of which is increased in the absence of ambient noise from a starting frequency (for example 20 Hz) to the cut-off frequency of the first high-pass filtering (here 180 Hz), and it is checked that at no time is n ^'obtained an increase in the gain of the amplifier by the first processed signal.

Prior to the first high-pass filtering of the first filtering subprocessing, the complex signal is filtered by an auxiliary high-pass filter having a cut-off frequency lower than that of the first high-pass filtering, and the comparison of level to decide on the possible blocking of the increase in the gain of the amplifier, between the signal sampled after this auxiliary filtering and the corresponding signal sampled after the first high-pass filtering; such an arrangement notably allows correct operation of the device when the vehicle is of a type emitting very low noise frequencies.

In order not to penalize the performance of such regulation, a decision is made, for each complex signal and useful signal taken present at the input of the corresponding processing chains, of the possible blocking of the increase in gain before having actually effected the possible increase in the gain of this amplifier taking into account the instantaneous signal / noise ratio.

According to another general feature of ^the invention, more particularly adapted to amplifiers having a plurality of input and output channels, are removed at any time the wanted signal on the channel having the level of the strongest signal, which allows particularly to overcome the problems of ^instability in the balance settings and fader. The invention also relates to an automatic gain control device for the amplifier of an electroacoustic system, comprising:

a means of capturing, such as a microphone, a complex signal containing a mixture of ambient noise and sound modulation coming from the loudspeakers.

means for acquiring a useful signal at the output of the gain-controlled amplifier.

- first means of processing the complex signal, incorporating first filtering means, capable of delivering a first processed signal.

- second means for processing the useful useful signal acquired, incorporating second filtering means, capable of delivering a second processed signal.

- main comparison means of the first and second processed signals, capable of delivering a main comparison signal; and

- control means able to control the increase or decrease in the gain of the amplifier as a function of the value of the main comparison signal; according to a first variant of the invention, each filtering means comprises a first high-pass to low-pass filtering stage cut-off frequency, as well as a second high-pass filtering stage with medium cut-off frequency and having attenuation at least of the order of 30 decibels at around 100 Hz; moreover, first auxiliary comparison means are provided, capable of comparing the respective levels of the signals sampled before and after the first high-pass filtering stage of the first filtering means, and of delivering a first auxiliary comparison signal having a first value when the level of the signal sampled before the first filtering stage is lower than the product of the level of the signal sampled after this stage, by a predetermined coefficient greater than 1, and a second value otherwise; second auxiliary comparison means are also provided, capable of comparing; εs respective levels of the signals taken before and after the first high-pass filtering stage of the second filter means, and to deliver a second comparison signal having a first value when the level of the signal sampled before the first high-pass filtering stage is greater than the product of the level of the signal sampled after this stage, by a predetermined coefficient greater than 1, and a second value otherwise; the control means are then able to block the increase in the gain of the amplifier when the first or the second auxiliary comparison means deliver their respective first values.

According to another variant of the invention, there is provided, in addition to a first high-pass filtering stage with low cut-off frequency, associated with a second high-pass filtering stage with medium cut-off frequency having attenuation of at least l about 30 decibels at approximately 100 Hz, a equalizing means set to prohibit, in the absence of noise, ^the increase in the gain of the amplifier for all pure signal (or useful) of frequency less than or equal to the cutoff frequency of the first high pass filter stage, ^the incorporation of such an equalizing means then makes ^it possible to dispense the second auxiliary comparator means and to retain only the first auxiliary comparator means; the control means are then able to block the increase of the gain of the ^amplifier, when the first comparing means auxiliaries deliver their first value.

The equalization means is advantageously a parametric equalizer with fixed correction preferably arranged downstream of the output of the gain-controlled amplifier. According to yet another variant of the invention, it is possible to dispense with an equalizer in the case where the electroacoustic system is self-supporting. It then suffices to provide, within each of the first and second filtering means, a first high-pass filtering stage at low cut-off frequency, as well as a second high-pass filtering stage at medium cut-off frequency and having attenuation at least of the order of 30 dB around 100 Hz; in combination, it is possible to dispense with the second auxiliary comparison means and keep only the first auxiliary comparison means which are capable of comparing the respective levels of the signals sampled before and after the first high-pass filtering stage of the first means of filtering, and to deliver a first auxiliary comparison signal having a first value when the level of the signal sampled before the first filtering stage is lower than the product of the level of the signal sampled after this stage, by a predetermined coefficient greater than 1 and a second value in the opposite case, the control means are then able to block ^the increase of the gain of the amplifier, when the first auxiliary comparator means outputs the first value.

The cut-off frequency of each first high-pass filtering stage is advantageously between approximately 150 Hz and approximately 200 Hz. Preferably 1 80 Hz, while the cut-off frequency of each second high-pass filtering stage can be between around 500 Hz and 2000 Hz, preferably 1000 Hz.

According to one embodiment of the invention, each of the first and second filter means comprises a high pass filter with a slope of 18 dB per octave approximately fom .ant the first filter stage high-pass, followed by ^a high pass filter having a slope of 6 dB per octave, fom.ant in combination with the slope filter 18 dB per octave, the second stage of high pass filtering. Each auxiliary comparison means may include a comparator, the input of which is connected to the input of the first corresponding high-pass filtering stage by a first resistor and to the output of this high-pass stage by a second resistor; the ratio of the first resistance to the second resistance defines said predetermined coefficient which is advantageously between approximately 1.33 and approximately 2.

According to one embodiment of the invention, the control means comprises a first transistor connecting the soπie first input processing means of a voltage follower whose output is connected to ^the control input of the gain-controlled amplifier, the base of which is connected to the output of each auxiliary comparison means by means forming a logical OR; a second transistor is also provided, disposed between ground and the input of the voltage follower, the base of which is controlled by the output of the main comparison means which include a comparator, the input of which is connected to the outputs of the two treatment by two resistors of unequal values and chosen ratio.

The equalizer is advantageously retractable which allows it to be incorporated or not into the device according to the invention. The presence of ^the equalizer in the device renders inoperative the second comparison means auxiliaries.

It is also particularly advantageous for the first filtering means to comprise an auxiliary high-pass filtering stage, disposed between the capture means and the corresponding first high-pass filtering stage, having a cut-off frequency lower than that of this first. high-pass filtering stage, preferably equal to 100 Hz, and the highest possible slope, for example 1 8 dB per octave.

According to another general feature of the invention, wherein ^the controlled-gain amplifier comprises a plurality of ^input channels and soπie. the acquisition means comprise discrimination means connected to the output channels of the gain-controlled amplifier, and capable of delivering at any time as output the signal from the output channel having the level foπ. According to an embodiment of the invention, in which the amplifier comprises a first pair of channels respectively connected to a first pair of loudspeakers, and a second pair of channels respectively connected to a second pair of loudspeakers, the discrimination means comprise: - four signal inputs respectively connected to the four channels,

- four controlled switching blocks, respectively connecting the four inputs to the line of the discrimination means,

a first means of comparing the sum of the levels of the channels of the first pair with the sum of the levels of the channels of the second pair,

- First control means able to open or close the two switches connected to the same pair according to the result of this first comparison. - second means for comparing the sum of the levels of a channel of the first pair and the homologous channel of the second pair, with the sum of the levels of the other channel of the first pair and of the homologous channel of the second pair .

- second control means suitable for opening or closing one of the two switches, not open under the action of the first control means; the only switch, finally closed, is the one connecting to the output of the discrimination means, the channel having the highest signal level.

Each switching block advantageously comprises two transistors whose collectors, common, are connected to the corresponding signal input and to the line of the discrimination means; the transmitters are connected to ground and the respective bases are controlled by the respective outputs of two comparators of the two comparison means. Other features and advantages of the invention will be ^the review of the EEA detailed description below of the appended drawings in which: Figure 1 is a block diagram of a general embodiment of a device according to the invention; Figure 2 illustrates in more detail a part of the device of the figure 1 ; Figure 3 illustrates an attenuation curve of a high pass filter used in the device of Figure 1; Figure 4 illustrates in more detail a part of an equalizer incorporated within the device of Figure 1; and FIG. 5 represents a more detailed block diagram of the discrimination means of the device of FIG. 1.

In FIG. 1, the reference 1 designates a gain-controlled amplifier, such as that sold by the company PHILIPS under the reference TDA 1 074. This amplifier 1 has four input and output channels, referenced 2, 3, 4 and 5 , as well as a gain control input 6 representing the pins "nine" and "ten" of the TDA 1074 component, joined together for this application. It is part of an electroacoustic audio frequency system, for example incorporated within a motor vehicle, and is connected, via the four channels, with an optional equalizer 7. and a means of amplification of power 8, four speakers 9 conventionally arranged in pairs at the front and rear of the vehicle.

The gain control chain of this amplifier 1 consists of two essential parts.

The first part comprises a capture means such as an ambient microphone 10, capable of picking up a complex signal containing a mixture of ambient noise and sound modulation coming from the loudspeakers. This microphone 10 is connected to ^input A of first means for Treatment 1 1 of a complex signal whose soπie D is connected ^by INTEM-ediate ^a resistor 12 to the inverting input ^of an operational amplifier 14 functioning in comparator, and whose non-inverting input is connected to ground. The second part of the control chain includes means for acquiring a useful or pure signal taken from the output of the gain-controlled amplifier 1, and incorporating here discrimination means 20, which will be discussed in more detail below. next on structure and function. Level adjustment means 21 are also provided, such as a potentiometer.

Soπie the potentiometer is connected to ^input A of second processing means 22 of the useful signal acquired, one soπie D is connected to the inverting input of the operational amplifier 14 by a resistor 13 less than the resistor 12 in a ratio R equal to the value of the resistor 12 on the resistor 13. The value of this ratio expressed in dB represents the margin of device security. The comparator line 14 is connected to the base of a transistor

PNP 16, operating as a switch, via a base resistor 15. The emitter of this transistor 16 is connected to ground and its collector is connected via a resistor 17 to the input d 'a voltage follower 1 9 whose line controls the gain of amplifier 1.

The reference 23 denotes a logarithmic amplifier whose function is to supply a voltage capable of controlling ^the gain controlled amplifier. Its input is connected to the soπie D of the first processing means, and its output is connected to the collector of the transistor 16, therefore to the input of the voltage follower 19, via the emitter-collector mesh of a transistor NPN 29. of which we will return in more detail below to the function, as well as by another resistor 30. This logarithmic amplifier 23 is conventionally composed of an operational amplifier 24 whose non-inverting input is connected to ground, and whose inverting input is attacked by a resistor 25.

This amplifier 24 is against feedback through a resistor 26 in parallel with two diodes in series 27 and 28, the cathode of the diode 28 being connected to ^the jug in input of the amplifier 24.

The basic principle of operation of this device is then as follows. When the level of the processed signal available at line D of the first processing means 1 1, is higher, taking account of the safety margin, at the level of the processed signal available at output D of the second processing means 22. that is to say -to say when the level of the complex signal processed is higher, taking into account the safety margin, at the level of the pure signal processed, the line of the comparator 14 is positive, which blocks the transistor 1 6 and allows, provided that the transistor 29 either passing, attacking the input of the voltage follower 19 with this processed signal present in line D of the first processing means, integrated according to a time constant (for example 4 seconds) defined by the product of the value of a capacitor 18 connected between the positive input of the voltage follower and ground, by the sum of the resistors 30 and 17. The increase in the gain of the gain-controlled amplifier 1 is then authorized.

Conversely, when the level of the pure signal processed, available in line D of the second processing means, is higher, taking into account the safety margin, than the level of the homologous signal available in line of the first processing means 1 1, the line of the comparator is negative, which turns transistor 16 on, allowing capacitor 18 to discharge through resistor 17, which leads to a decrease in the gain of amplifier 1.

The base of transistor 29 is controlled, via a resistive bridge 31 and 32 which is also connected to the positive supply of the device, by the anode voltage of two diodes 33 and 36, the respective cathodes of which are connected to the lines of two comparators 34 and 37 appearing respectively in first and second auxiliary comparison means 35 and 38, the structure of which will now be described in more detail with particular reference to FIG. 2.

In this figure, it can be seen that the first auxiliary comparison means 35 comprise two rectifiers without threshold 54 and 55 respectively connected to the two lines B and C of the first processing means 1 1. More particularly, the rectifier without threshold 54. connected to the soπie B, and providing a positively rectified signal, conventionally comprises an operational amplifier whose positive input is connected to ground, and whose negative input, driven by a capacitor in series with a resistance, is counter-reacted by a first diode. Another diode is connected in series with the first, the two diodes being connected in parallel to the terminals of a resistor and ^a filter capacitor. The direction of adjustment is provided by the diode mounting direction (anode connected to the inverting input of ^the amplifier).

The rectifier without threshold 55, supplying a rectified - negative voltage, has a structure similar to that of the rectifier without threshold 54 in the direction of mounting near the two diodes.

The outputs of these two thresholdless rectifiers 54 and 55 are respectively connected to the positive input of comparator 34 by the intermediary of two resistors 56 and 57. A person skilled in the art therefore notices here that the first auxiliary comparison means, and more particularly the comparator 34, are capable of comparing the level of the signal present at line B at the level of the signal present in line C of the first processing means 1 1, to within a predetermined coefficient. In other words, the comparator 34 compares the level of the signal in B to the product of the level of the signal in C by a predetermined coefficient, defined by the ratio of the resistance 56 to the resistor 57. This ratio is advantageously chosen between 1.33 approximately and approximately 2. for reasons which will be explained in more detail below.

The point B materializes here the output of an auxiliary high-pass filtering stage 39 whose input A is connected to the microphone 10. This auxiliary stage co poπe as input a capacitor 47 in series with a resistor 48 connected to ground. The non-inverting input of an operational amplifier 40 is connected on the one hand to ground by a resistor 41, and on the other hand to the common point of the resistor 48 and of the capacitor 47 by means of two capacitors in series 42 and 46. the inverting input of this amplifier 40 is connected to the common point of the two capacitors 42 and 46 via a resistor 43 and to ground ^through a resistor 44. the amplifier is contraindicated reacted by a resistor 45. The assembly forms a high-pass filter having a cut-off frequency of 50 Hz to 100 Hz with a slope of 18 dB / octave. The resistive bridge 4-4 and 45 provides a gain of 50 to this auxiliary stage 39. which makes it possible to adapt to a correct level the level 'of the signal received from the microphone 10.

The B soπie auxiliary dε cεt filtering stage high-pass 39, also performs ^the input ^of a first filter stage 49. highpass without gain, having a cutoff frequency of 1 80Hz εt a slope of 18 dB / octave. The structure of this first high-pass filtering stage

49 is identical to that of the operational amplifier circuit of the auxiliary stage 39.

The line C of this first high-pass filtering stage 49 is connected to the input of a second high-pass filter 50. having a significant cutoff frequency equal to 1000 Hz and a pεntε of 6 dB / octave. A such a filter is produced conventionally by a capacitor 52 of 4.7 nF connected in series to ground with a resistor 5 1 of 47 kΩ.

The combination of the filter stage 49 and the filter stage 50 forms a high-pass filter having a cut-off frequency of the order of 1000 Hz as well as an AC attenuation curve (FIG. 3) presenting a attenuation of the order of - 40 dB at 100 Hz, and of the order of - 20 dB at 200 Hz.

The signal available at the output of the filter 50 is then detected at the peak by a positive peak detector 53, the line D of which represents that of the first processing means 11.

The second processing means 22 are similar to the first processing means 1 1 with only one difference, namely that the auxiliary filtering stage having a cut-off frequency of 100 Hz is eliminated since it is unnecessary. Points A and B are therefore confused. The output of the second filtering means this time attacks a negative peak detector.

The second auxiliary comparison means 38 are analogous to the first auxiliary comparison means 35 with the difference that the point E is this time connected to the negative input of comparator 37 whose positive input is connected to ground (FIG. 1 ).

According to a variant embodiment of the device according to the invention, it is provided to insert downstream of the controlled gain amplifier 1, ^the equalizer 7 is. for this purpose, advantageously produced in the form of a retractable box, which the user can plug into his car radio box. The insertion of this equalizer 7 has the effect of connecting the negative feed point to the inverting input of comparator 37. which forces its output to a positive value, which. in combination with the diode 36 inhibits the second auxiliary comparison means 38. This equalizer, preferably parametric and fixed correction, is illustrated schematically in Figure 4. It is arranged ⁱⁿ a general way to avoid, in the absence of noise, a switching of the comparator 14 in the direction of increasing the gain of the ^amplifier for any signal less pure frequency or equal to the cutoff frequency of the first filter stage pass high 49, that is to say for any frequency less than or equal to 180 Hz in the present case.

The equalizer comprises between the input F and the line G of each channel, an operational amplifier 58 mounted as an inverter, of unity gain determined by the two resistors 60 and 59 of equal values, for example 22 kΩ. One or more equalization cells 61 are further provided, connected to the terminals of the inverter stage, and comprising an operational amplifier 66 mounted as a bandpass filter with adjustable frequency by resistors 68 of equal values, between 470 Ω and 470 kΩ and adjusted according to the installation. The amplifier 66 is supplied by a divider bridge constituted by the resistors 63 and 64. Furthermore, two coupled resistors 62 of variable value, connected in series to the terminals of the input F and of the line G. constitute a potentiometric assembly which peπnet to vary the gain of the bandpass filter as a function of the ratio of the two resistors 62, while the bandwidth is adjusted by the ratio of the resistor 63 to the resistor 64.

In the present case, the sum of the two resistors 62 is taken equal to 47 kΩ. while resistance 63 has a value less than 2

MΩ, also adjusted according to the installation.

We will now refer more particularly to FIG. 5 to describe in more detail the means of discrimination 20.

The input H of the discrimination means 20 is connected to the four channels 2.3.4.5 which will be considered here in pairs, namely, for example, a first pair of channels 3 and 4 respectively connected to the two rear speakers and a second pair of channels 2 and 5 respectively connected to the two front speakers.

The four channels are connected to the output I of these discrimination means by four controlled switching blocks. Two of these blocks respectively consist of two identical transistors

• PNP or NPN bearing the references 85. 86. 93 and 94. The two other blocks are respectively made up of two complementary transistors PNP and NPN 87.88; 97.98. The transmitters of each of the two transistors forming a switching block are connected to the ground, while the two collectors are connected on the one hand to a channel by a resistor 101, and to the line 1 by a resistor 102.

A switching block will be considered to be open, that is to say not connecting the channel corresponding to the line, when one of the two transistors constituting it is on. The switching block will be considered closed, that is to say connecting the channel corresponding to the line when the two transistors constituting it are blocked.

The inverting input of an operational amplifier 71, mounted as a comparator, is connected via a summing rectifier 75, composed of two resistors 81 and 82, respectively connected to the first pair of channels 3 and 4, and followed a rectification cell similar to that of the rectifier 54.

The non-inverting input of the operational amplifier 71 is connected by another summing rectifier 76. 83 and 84, of structure similar to that of the summing rectifier 75, to the second pair of channels 2 and 5.

A second operational amplifier 72, analogous to the amplifier 71, is also mounted as a comparator. Its negative input is also connected by the intermediary of a summing rectifier 73,77.78 to a channel of the first pair and to a channel of the second pair, in other words, for example, here, to the front channel left 5 and to the left rear channel 4. The positive input of this comparator 72 is connected by another summing rectifier 74,79.80. respectively to the other two channels of the two pairs, that is to say here to the right front and rear channels.

Soπie the comparator 71 is connected, ^via four base resistors 95.89.91 and 100 to the respective bases of the four transistors 85.93.87 and 97 of the four switching units, while the comparator 72 is soπie reliéε by ^' intermediate of the four basic resistors 96.99.90 and 92. to the respective bases of the four associated transistors 94.98.86 and 88 of these four switching blocks.

We will now describe in detail the operation of the device according to the invention. Curve ^AC attenuation of the second filter stage pass high was chosen to be close to that of the spectral envelope of human speech in its low side (frequency less than 1000 Hz), because it was observed that it corresponded statistically to the most probable and most useful envelope of the signal. The adoption of such a curve makes it possible, on the one hand, to take account of effectively annoying car noises in order to mask in particular human speech, or any musical signal having a comparable spectral envelope, and. on the other hand, to greatly attenuate the low frequency noises which are not really annoying as regards the masking of the signal.

If the attenuation curve chosen here contributes to obtaining very good performance of the device, it has been observed that it is generally possible to adopt a high-pass filter having a cut-off frequency of between approximately 500 Hz and around 2000 Hz, and attenuation at least of the order of 30 db around

100 Hz.

The noise emitted by a car is generally rich in the low frequencies, the adoption of an attenuation curve according to ^the invention contributes to a decrease. spectral density of noise and therefore to a reduced difference between the level of the noisy signal at line D of the first processing means 11 and the level of the non-noisy signal at output D of the second processing means 22. If one wishes to avoid a untimely tilting of the device in the direction of a runaway, one solution would be to adopt a margin of safety, relatively impoπante. However, this is not satisfactory because it would lead to the comparator 14 controlling the transistor 16 in order to authorize the increase in the gain of the amplifier, only in the presence of a noise level such that in any case ^the auditor would pierce any more the signal emitted by the loudspeaker before tipping system.

Also, it was observed that it was advisable to adopt a safety margin between approximately 1.5 dB and approximately 4 dB, preferably around 2 dB.

To perform ^calibration of the device, the procedure is as follows. We introduce a pink noise modulation in the gain-controlled amplifier and, using the potentiometer 21, a level adjustment is effected so as to obtain respective output voltages at points D of the first processing means 11 and the second processing means 22, which are equal in ultimate value. Given this setting and the safety margin chosen, the comparator

14 will only trigger an increase in gain when the voltage at point D of the first processing means is 2 dB (safety margin) greater than that available at point D of the second processing means 22. A rapid calculation shows that '' at this time, the ambient sound pressure paπielle ambient noise is 2 dB lower than that of the sound modulation from the speaker, an improvement of 13 dB compared to the known device without equalizer and 7 dB for those with 'equalizer, the performance of this type of system in terms of overcompensation.

Those skilled in the art will understand that with such a safety margin, the risks of the device running away during normal operation with a noisy signal are significant. Also, the first and second auxiliary comparison means 35 and 38 contribute, in the first alternative embodiment of the device, to avoid such runaway, in the same way as the first auxiliary comparison means 35 in combination with the equalizer 7 , in the second variant, and only the first auxiliary comparison means 35, in the third variant of the invention. concerning a freestanding electroacoustic chain.

The first auxiliary comparison means 35 compare the level of the noisy signal sampled before the first high-pass filtering stage 49 with the level of this same signal sampled in line from this filtering stage 49, taking account of the resistor 56 and 57 Thus, when the level of the signal taken at point B is less than the product of the level of the corresponding signal taken at point C by the comparison of resistors 56 of 57. the voltage delivered by comparator 34 is negative, which blocks the transistor 29 and samples the voltage by the voltage follower 1 9 if transistor 16 is not on. The follower 19 keeps thanks to the capacitor 18 sa * ^• *.

value and prevents the increase in the gain of amplifier 1 and consequently a runaway of the device. It has in fact been observed that the transistor 29 should be blocked when the level of the signal at B is not sufficiently higher than the level of the signal at C, because in this case, the processed signal delivered by the first processing means 1 1 may not be representative of car noise.

In the opposite case, that is to say when the level of the signal taken at point B is greater than the product of the level of the signal taken at point C by the resistor 56/57. it is therefore likely that the signal delivered by the first processing means is actually tainted by a car noise. The transistor 29 is then in its on state obtained by the combination of the positive line of the comparator 34 and the diode 33.

The choice of the ratio of the resistances 56/57 (between approximately 1.33 and approximately 2) is an experimental compromise between, in a pan, an autoblocking of the system when the ambient noise contains high frequencies, and. ^{on the} other side. a difficulty in blocking resulting in overcompensation harmful to the proper functioning of the device.

Moreover, the time constant defining the reaction rate of the transistor 29 (defined by the value of the resistor and capacitor of threshold without rectifiers 54 and 55 and taken here equal to ^the order of 1/10 of a second), is very low compared to the time constant defined by the resistors 30, 17 and the capacitor 18 (of the order of 4 seconds). Also, although the transistors 16 and 29 react at the same speed, seulε a fraction of the voltage supplied by ^the logarithmic amplifier 23 applied sεra the voltage follower 19. the time that the transistor 29 is blocked. In other words, a decision is made here on the possible blocking of the increase in the gain of the amplifier via the transistor 29. before the switching of the comparator 14 has possibly commanded this increase.

With regard to the choice of the cut-off frequency of the first high-pass filtering stage 49. it has been observed that this frequency should be lower than the fundamental of a low male voice, and higher than the fundamental the highest of a car noise. A satisfactory solution has led to the choice of a frequency lying around 1 80 Hz although this range can be extended between approximately 1 50 Hz and approximately 200 Hz.

Furthermore, although it is possible to dispense with the high-pass filtering auxiliary stage 39, having a cut-off frequency of the order of 50 Hz (but possibly up to approximately 100 Hz), it has been deemed preferable to take the signal at point B, that is to say before the filtering stage 49, but after this auxiliary stage 39, so as to filter out very low noise frequencies which can occur in certain areas types of vehicles. In that case. in the absence of such an auxiliary filtering stage 39. it would be advisable to increase the ratio of the resistors 56 to 57 in a too impoπant manner which would harm the proper functioning of the device as has been explained above.

The choice of the cut-off frequency of stage 39 compared to that of stage 49 also influences the choice of the predetermined coefficient defined by the ratio of resistors 56/57. Thus, for a frequency ratio of the order of 1/2, we will preferably choose a coefficient of 1.33 while for a ratio of the order of 1/4, we will prefer a coefficient of the order of 2 . the deuxièmεs auxiliary comparison means 38 allow, in the first embodiment variant of the device, to prevent a runaway of the system, ^that is to dirε unε augmεmation iπéversible the gain of the ^amplifier 1, on the useful signal frequency less than or equal to the cutoff frequency of the first high-pass filtering stage 49, that is to say for a frequency less than or equal to 180 Hz.

In fact. In this case there is a signal level sampled at point B of the second processing means 22. greater than the product of the corresponding signal level sampled at point C by the predetermined coefficient equal to the ratio of resistances 56/57, the comparator 37, in combination with diode 36. blocks transistor 29, which leads the voltage follower to keep its previous control value.

It is possible to ^release itself from these dεuxièmes auxiliary comparator means 38 by inserting ^the equalizer 7 in the electro-acoustic system. in the embodiment described, makes the second auxiliary comparison means 38 inoperative. In fact, the equalizer 7 is adjusted to a chosen setting value such that, for no pure signal frequency lower than the cutoff frequency of the first high-pass filtering, it is that is to say about 180 Hz, the level of the voltage coming from the microphone and reaching the main comparator 14 is higher than the voltage coming from the second processing means, and reaching this same comparator, taking account of course of the margin of security adopted. In other words, this setting value is chosen so as to avoid, in the absence of noise, a tilting of the comparator in the direction of an increase in the gain of the amplifier for any pure signal frequency less than 180 Hz. Thus, a person skilled in the art will understand that, with such an adjustment, any runaway of the system is avoided during operation on any useful signal less than 180 Hz. The equalizer ε is preferably placed downstream of the amplifier 1, because it contributes both to the stability of the device and to the quality of hearing. Nevertheless, it can be placed elsewhere, for example just before the second processing means 22.

It is also possible, in another variant of the invention, to do without such an equalizer. This is possible in the case where the electroacoustic system is freestanding below the cutoff frequency (180 Hz) of the first high pass filtering stage. Within the meaning of this invention. an electroacoustic system is said to be autostable, below this cut-off frequency, if, after the pink noise calibration phase, a sinusoidal signal of adjustable frequency is injected as pure signal modulation, the frequency of which is increased the ^absence of ambient noise from a dépaπ frequency by exemplε 20 Hz to the cutoff fréquεncε (1 80Hz), and if at any time, the main comparator 14 does not swing in the direction ^of an increase in the amplifier gain.

In such a case. the first auxiliary comparison means 35 are in themselves sufficient to ensure the stability of the system and to solve the problem posed in combination with the filters of the first processing means 11. It should also be noted that the combination of the stages of filtering 39 and 49 also makes it possible to discriminate between the noise produced by the vehicle and the passenger conversation, which prevents this conversation from being considered as noise which would lead to an increase in gain in order to cover this voice. In addition, the use of peak detectors makes it possible to reduce the disparity in reaction of the device vis-à-vis pure sustained notes resulting in a very low signal / noise ratio and passages containing a lot of transients in which the signal ratio / noise is very foπ. In order to avoid inadvertent gains in gain on “almost white” dεs, suitable means 103 are provided which can be incorporated into each of the variants mentioned above. These means 103 (FIG. 1) mainly consist of a rectification cell 104 analogous to rectifier 54. responsible for rectifying the signal coming from the discrimination means 20, and which directly supplies, in one section, the positive input of an operational amplifier 105 mounted as a comparator, and on the other hand, its negative input via a voltage divider bridge formed by resistors 108 and 1 10 of equal sensiblemεnt values. A resistor 109, connected between the negative input 2 of the comparator 105 and the positive supply point, of very great value compared to that of the resistors 108 and 1 10. forms with the resistor 108 a voltage divider in a ratio of 40 to 50 dB.

A capacitor 107 is connected between the negative input of comparator 105 and the ground, and the line of comparator 105 is connected to the base resistance 32 of transistor 29 via a diode 106.

These means 103 operate in the following manner: when no sound modulation is emitted, which for example corresponds to the start of a disc, the voltage from the rectifier 104 is zero, and. by the resistor 109 connected to the most of the power supply,

* the inverting input of amplifier '105 is slightly positive, which makes its line negatized. and, ^through the diode 106 turns off transistor 29 and prevents the rise of the gain in the absence of sound modulation. As soon as the signal rises, the comparator line 101 becomes positive, which makes the transistor 29 pass. In steady state, due to the resistive bridge 106, 104, the voltage at the positive input of comparator 105 is greater than that of the negative input, which has the consequence of maintain the line of the comparator 105 positive and the transistor 29 in its on state.

If the signal drops rapidly by more than 6 dB, the positive input of comparator 105 instantly follows this decrease, but not its negative input, whose voltage drop is delayed by the capacitor 107. The line of comparator 105 becomes negative again, this which blocks the transistor 29. On the other hand, if the useful signal drops progressively, the capacitor 107 discharges quickly enough so that the voltage at the negative input of comparator 105 does not become greater than that of the input positive, which allows the transistor 29 to remain on and the gain to rise when the signal drops without interrupting suddenly, which preserves the main purpose of the whole device.

We have therefore just seen that the means 103 are capable of blocking the increase in gain in the absence of sound modulation or in the presence of a decrease in useful signal greater than a predetermined decrease (in the present case a decrease in more than 6 dB and faster than the discharge of the capacitor 107). The values of the various elements of these means 103 result from experimental choices. However, good results were obtained with a capacitor 107 ^to a value of 10 microfarads. a resistor 109 of the order of 10 MΩ. and resistors 108 and 1 1 0 of the order of 22 kΩ.

In order not to cause instability of the system when adjusting the balance and / or fadεr dε chain, the first comparator 71 of the discrimination means 20. compares the sum of the levels of channels 3 and 4 dε the first pair with the sum of the levels of channels 2 and 5 of the second pair and controls the opening of the two switches connected to the pair providing the lowest sum, while the second comparator 72 compares the sum of the levels of a channel of the first pair and the homologous channel of the second pair, with the sum of the levels of the other channel of the first pair and of the homologous channel of the second pair, and controls the opening of that of the two switches, not open under the action of the first control means, and connected to the channel whose level has contributed to lead to the lowest sum.

5 Thus, the only switch which is finally closed, at all times connects the line I of the discrimination means 20 to the channel having the highest signal level.

Concretely, if the fader is activated to favor the two front channels, the line of comparator 71 is positive, which makes

10 passing the transistors 93 and 87, consequently inhibiting the two rear channels 3 and 4 (open switches).

If the balance is actuated to favor the left channels 4 and 5, the output of the comparator 72 is negative, which makes the transistors 86 and 88 on, consequently inhibiting the two right channels. The

15 comparator 72 therefore opened the switch 86,85, which had not been opened under the action of the comparator 71. Thus, only the front left channel 5 is connected to the output I of the discrimination means, the transistors 97 and 98 remaining blocked by the action of comparators 71 and 72. .- Proceeding in this way, when one deviates from the midpoint of the balance and fader settings, the signal from amplifier 1 remains constant and that coming from microphone 10 can only go down. We can then get under compensation, which is much preferable to runaway.

25 Of course, other means can be imagined to select the most powerful channel, for example an optical location on mechanical potentiometers or else the reading of memories in the case of electronic potentiometers.

Likewise, although all the means of the present invention have

30 have been carried out here in an analog manner, one could conceive of carrying them out at least in digital form, in particular with regard to filtering, by means for example of a processor for processing the signal.

Claims

28

1. Method for automatically controlling the gain of an amplifier of an electroacoustic system, in which: a) a first processing (1 1) is carried out, comprising a first filtering subcontracting, on a complex signal containing a mixture of ambient noise and sound modulation from the speakers, to obtain a first processed signal. b) a second processing (22) is carried out, comprising a second filtering subprocessing, on a useful signal taken at the output of the gain-controlled amplifier (1). to get a second processed signal. c) lε dε gain is regulated amplifier (1) based ^on a level comparison between the two traiiés signals, characterized in that each filter sub-processing comprises a first high-pass filtering (49) Low cut-off frequency, as well as a second high-pass filtering (49.50) with medium cut-off frequency and having attenuation at least of the order of 30 dB around 100 Hz. the increase in the gain of the amplifier (1) is blocked when the level of the signal sampled before the first high-pass filtering (39) of the first filtering subcontracting is less than the product of the level of the corresponding signal sampled after this first high-pass filtering (49). by a predetermined coefficient (56/57) greater than 1. or when the level of the signal pre-ee before the first high-pass filtering (49) of the second filtering sub-processing is greater than the product of the level of the corresponding signal pre-ees after this first high-pass filtering, by a predefined coefficient greater than 1.

2. Method for automatically controlling the gain of an amplifier of an electroacoustic system, in which: a) a first processing (1 1) is carried out, comprising a first filtering sub-processing, on a complex signal containing a mixture of ambient noise and sound modulation coming from loudspeakers, to obtain a first processed signal, b) a second processing is carried out (22). with a second filtering sub-processing, on a useful signal sampled in loudness from the gain-controlled amplifier (1), to obtain a second processed signal, c) the gain of the amplifier (1) is regulated as a function of a level comparison between the two processed signals, characterized in that each filtering subprocessing comprises a first high-pass filtering (49) at low cut-off frequency as well as a second high-pass filtering at medium cut-off frequency and having attenuation of at least about 30 dB around 100 Hz, by the fact that an equalizer is provided which is adjusted (7) so as to prohibit the increase in the gain of the amplifier by any useful signal of frequency less than or equal to the cutoff frequency of the first high-pass filtering (49), in the absence of noise, and by the fact that the increase in the gain of the amplifier (1 ) when the level of the signal sampled before the first filtering high pass (49) of the first filtering subprocessing is less than the product of the level of the corresponding signal sampled after this first high pass filtering (49) by a predetermined coefficient (56/57) greater than 1.

3. A method of automatically controlling the gain of an amplifier ^to an electro-acoustic system, wherein: a) effecting a first processing (1 1). comprising a first filtering sub-processing, on a complex signal containing a mixture of ambient noise and sound modulation coming from the loudspeakers, to obtain a first processed signal, b) a second processing (22) is carried out, comprising a second filtering subprocessing, on a useful signal taken at the output of the gain-controlled amplifier (1). to get a second processed signal. c) controls the gain of the amplifier (1) based ^on a level comparison between the two ^'processed signals. characterized by the fact that each filtering sub-treatment comprises a first high-pass filter (49) at low cut-off frequency, as well as a second high-pass filter (49.50) at medium cut-off frequency and having attenuation at least of the order of 30 dB around 100 Hz, by the fact that the electroacoustic system is self-supporting below the cut-off frequency of the first high-pass filtering, and by the fact that the increase in the gain of the amplifier (1) is blocked when the level of the signal sampled before the first high-pass filtering (39) of the first filtering subprocessing, is less than the product of the level of the coπesponding signal sampled after this first high-pass filtering (49), by a predetermined coefficient (56/57) greater than 1

4. Method according to one of claims 1 to 3, characterized in that it compoπe a calibration phase in which a pink noise modulation is injected at the input of the amplifier, and a level adjustment is made prohibiting increasing the gain of the amplifier by the first signal processed, taking into account a safety margin chosen between approximately 1.5 dB and approximately 6 dB, preferably 2 dB.

5. Method according to any one of the preceding claims, characterized in that a second high-pass filtering (49.50) is chosen having an attenuation curve (CA) having an attenuation of the order of - 40 dB at 100 Hz and around - 20 dB at 200 Hz.

6. Method according to one of the preceding claims, characterized in that prior to the first high-pass filtering (49) of the first filtering sub-processing, the complex signal is filtered (39) by an auxiliary high-pass filter having a cut-off frequency lower than that of the first high-pass filtering (49). and in that one carries out the comparison level to determine the possible blocking of increasing the gain of the ^amplifier, between the signal collected after the auxiliary filter (39) and the corresponding signal taken after the first pass filtering -high (49).

7. Method according to one of the preceding claims, characterized in that for each complex signal and useful signal sampled, present at the input of the corresponding processing chains, it is decided whether to block the increase in gain ( 29) before having actually carried out the possible increase (16) of this gain.

8. Method according to one of claims 1 to 7, wherein the amplifier compoπe a plurality of input and output channels, characterized in that the useful signal is taken at any time from the channel having the highest signal level.

9. Automatic gain control device for an amplifier of an electroacoustic system, including:

a means for capturing (10) a complex signal containing a mixture of ambient noise and sound modulation coming from the loudspeakers,

means of acquisition (20) of a useful signal at the output of the amplifier,

- first processing means (11) for the complex signal, incorporating first filtering means, capable of delivering a first processed signal,

- second processing means (22) of the useful signal acquired, incorporating second filtering means, capable of delivering a second processed signal,

- main comparison means (14) of the first and second processed signals, capable of delivering a main comparison signal, and control means (16) capable of controlling the increase or decrease in the gain of the amplifier as a function of the value of the main comparison signal, characterized in that each filtering means comprises a first high-pass filtering stage (49) at low cut-off frequency, as well as a second high-pass filtering stage at medium frequency cutoff (49.50) and having an attenuation of at least about 30 dB around 100 Hz, by the fact that there is further provided first auxiliary comparison means (35), capable of comparing the respective levels of the signals sampled before and after the first high-pass filtering stage (49) of the first filtering means, and in delivering a first auxiliary comparison signal having a first value when the level of the signal pr high before the first filter stage (49) is lower than the product of the level of the signal taken after this stage (49), by a predetermined coefficient (56/57) greater than 1, and a second value in the opposite case,

- second auxiliary comparison means (38), capable of comparing the respective levels of the signals sampled before and after the first high-pass filtering stage (49) of the second filter means, and of delivering a second comparison signal having a first value when the level of the signal sampled before the first filtering stage (49) is greater than the product of the level of the signal sampled after this stage, by a predetermined coefficient (56/57) greater than 1, and a second value in the case on the contrary, and by the fact that the control means (29) are capable of blocking the increase in the gain of the amplifier when the first (35) or second (38) auxiliary comparison means deliver their first values.

10. Device for automatically controlling the gain of an amplifier of an electroacoustic system, comprising:

means of acquisition (20) of a useful signal in line with the amplifier,

- first processing means (1 1) for the complex signal, incorporating first filtering means, capable of delivering a first processed signal,

- 'main comparison means (14) of the first and second processed signals, capable of delivering a main comparison signal, and control means (16) capable of controlling the increase or decrease in the gain of the amplifier in function of the value of the main comparison signal, characterized in that each filtering means comprises a first high-pass filtering stage (49) with low cut-off frequency, as well as a second high-pass filtering stage (49.50) with medium cut-off frequency and having attenuation at least of the order of 30 dB around 100 Hz, by the fact that there is further provided an equalization means (7) adjusted to prohibit the increase in the gain of the amplifier for any useful signal of frequency less than or equal to the cut-off frequency of the first high-pass filtering stage (49), in the absence of noise, by the fact that first auxiliary comparison means (35) are provided, capable of comparing the respective levels of the signals sampled before and after the first high-pass filtering stage (49) of the first filtering means, and delivering a first auxiliary comparison signal having a first value when the level of the signal sampled before the first filtering stage (49) is lower than the product of the level pre signal after this stage (49), by a predetermined coefficient (56/57) greater than 1, and a second value otherwise, and by the fact that the control means are capable of blocking the increase in gain of l amplifier when the first auxiliary comparison means (35) deliver their first value.

1 1. Device according to claim 10, characterized in that the equalization means is a parametric equalizer with fixed correction, arranged downstream of the line of the gain-controlled amplifier (1).

12. Device according to claim 1 1 taken in combination with claim 9, characterized by the fact that the equalizer (7) is retractable, and by the fact that the presence of the equalizer (7) in the device renders the second auxiliary comparison means (38).

13. Device for automatically controlling the gain of an amplifier of an electroacoustic system, comprising:

means of acquisition (20) of a useful signal at the output of the amplifier, - first processing means (1 1) for the complex signal, incorporating first filtering means, capable of delivering a first processed signal,

- main comparison means (14) of the first and second processed signals, capable of delivering a main comparison signal, and - control means (16) capable of controlling the increase or decrease in the gain of the amplifier by function of the value of the main comparison signal, characterized in that each filtering means comprises a first high-pass filtering stage (49) at low cut-off frequency, as well as a second high-pass filtering stage at medium cut-off frequency (49.50) and having attenuation of at least 30 dB at around 100 Hz, due to the fact that the electroacoustic system is self-supporting below the cut-off frequency of the first filtering stage -high (49), by the fact that first auxiliary comparison means (35) are provided, capable of comparing the respective levels of the signals sampled before and after the first high-pass filtering stage (49) of the first s filtering means, and to deliver a first auxiliary comparison signal having a first value when the level of the signal sampled before the first filtering stage (49) is lower than the product of the level of the signal sampled after this stage (49), by a predetermined coefficient (56/57) greater than 1, and a second value in the opposite case, and by the fact that the control means are capable of blocking the increase in the gain of the gain-controlled amplifier when the first means auxiliary comparison (35) deliver their first value.

14. Device according to one of claims 9 to 13, characterized in that the cut-off frequency of each first stage of high pass filtering (49) is between about 150 Hz and about 200 Hz, while the cutoff frequency of each second high pass filtering stage (49/50) is between about 500 Hz and about 2000 Hz, from preferably 1000 Hz. 15. Device according to claim 14, characterized in that each of the first and second filtering means comprises a high-pass filter (49) having a slope of approximately 18 dB per octave, forming the first filtering stage high pass (49), followed by a high pass filter (50) having a slope of 6 dB per octave, forming in combination with the slope filter 18 dB per octave (49), the second pass filter stage high (49/50).

16. Device according to one of claims 9 to 15, characterized in that each auxiliary comparison means comprises a comparator (34,37) whose input is connected to the input (B) of the first filtering stage pass -high corresponding, by a first resistance

(56) and at the line (C) of this high-pass stage (49) by a second resistor (57), the comparison of the first resistor (56) with the second resistor (57) defining said predetermined coefficient.

17. Device according to one of claims 9 to 16, characterized in that the coefficient is between approximately 1.33 and approximately

2.

18. Device according to one of claims 9 to 17, characterized in that the main comparison means (14) compare the first and second signals processed taking into account a chosen safety margin, and in that the means command component:

- a first transistor (29) connecting the output (D) of the first processing means (11) to the input of a voltage follower (19) whose output is connected to the control input of the amplifier at controlled gain (1), and the base of this transistor being connected to the output of each auxiliary comparison means (35,38) by means of a logical OR means (33,36), and

- a second transistor (16), disposed between ground and the input of the follower (19), the base of which is controlled by the output of the main comparison means comprising a comparator (14) whose input is connected to the lines of the two processing means (11, 22) by two unequal resistances (12, 13), of chosen ratio, defining said safety margin.

19. Device according to one of the preceding claims, characterized in that the first filtering means comprise an auxiliary high-pass filtering stage (39), disposed before the first high-pass filtering stage (49), having a cut-off frequency lower than that of the first high-pass filtering stage and the highest possible slope. 20. Device according to one of claims 9 to 19, characterized in that it comprises means (103) capable of blocking the increase in gain in the absence of sound modulation or in the presence of a signal decrease useful above a predetermined decrease. 21. Device according to claim 20 taken in combination with claim 17, characterized in that the means (103) compoπent a comparator (105) whose line is connected to the base of the first transistor (29) via d a diode (106), the non-inverting input of which is connected to the line of a rectifying cell (104), and the inves input of which is firstly grounded to ground by means of a capacitor (107), secondly, at the bottom of the rectification cell (104) via a resistive divider bridge (108, 1 10), and, thirdly, at a supply point by a resistor (109) of very great value compared to that of the resistors of the resistive bridge (108, 1 10).

22. Device according to one of claims 9 to 21, characterized in that the gain-controlled amplifier (1) comprises several input and output channels (2,3,4,5), and in that that the acquisition means comprise discrimination means (20) connected to the output channels of the gain-controlled amplifier, and capable of delivering at any time in output (I) the signal from the output channel having the most foπ level.

23. Device according to claim 22, characterized in that the amplifier compoπe a first pair of channels respectively connected to a first pair of speakers, a second pair of channels respectively connected to a second pair of loudspeakers, by the fact that the discrimination means (20) comprise:

- four signal inputs respectively connected to the four channels (2,3,4,5), - four controlled switching blocks, respectively connecting the four inputs to the loudness of the discrimination means,

- a first means of comparison (71) of the sum of the levels of the channels of the first pair with the sum of the levels of the channels of the second pair, - first control means able to open or close the two switches connected to the same pair according to the result of this first comparison,

- second comparison means (72) of the sum of the levels of a channel of the first pair and of the channel homologous to the second pair, with the sum of the levels of the other channel of the first pair and of the channel homologous of the second pair, and

- second control means suitable for opening or closing one of the two switches, not open under the action of the first control means. 24. Device according to claim 23, characterized in that each switching block comprises two transistors, the common collectors of which are connected to the coπesponding signal input and the line of the discrimination means, the transmitters of which are connected to the mass, and whose respective bases are controlled by the respective outputs of the two comparators (71, 72).