EP1211668A1 - Active acoustic reflector - Google Patents

Abstract

The sound wave active reflecting mechanism has sound detectors and pre amplifiers with power amplifiers amplifying sound to loudspeakers. The distance between each sound detector and loudspeaker is set between 20 cm and 2m. There is an amplifier control circuit producing the loudspeakers and sound detectors. The directivity of the detectors and loudspeaker assemblies is controlled together with reverberation filtering.

Description

The present invention relates to active correction of the acoustics of performance halls, auditoriums, auditoriums conferences and amphitheatres, etc. The idea of the reflector active sound (where “active shutter”) derives from that of traditional (passive) louvers, sort of reflector usually placed above the proscenium to return towards the audience of the sound energy coming from the scene. This "early" energy, arriving soon after the "sound direct "(sound propagating directly from the source to listeners), reinforces intelligibility and clarity messages perceived by listeners. When the active sound reflectors are placed on the sides of a room, the reinforcement of early energy also results in a pleasant feeling of enlargement of the source on scene. Through the use of techniques of electroacoustics (microphones, filters, amplifiers, active speakers) have a significantly higher efficiency than louvers liabilities. Their dimensions can be much more and their behavior (e.g. intensity with which they reflect the sound) can be adjusted by the user according to the requirements of each type of show. A set of active sound reflectors will also increase the reverberation of the room in which they are placed, with or without the help of filters electronic reverberators.

• Renfort du son direct, grâce à des réflecteurs sonores actifs placés au dessus de l'avant scène, ou à proximité du cadre de scène.
• Conque de scène active. Amélioration de l'acoustique sur scène grâce à des réflecteurs sonores placés autour des musiciens.
• Augmentation de la durée de réverbération d'une salle grâce à un ensemble de réflecteurs sonores actifs.
• Création d'un effet de salle dans un spectacle en extérieur.
• Renforcement local du niveau sonore dans la salle, grâce à l'utilisation d'un ou plusieurs réflecteurs sonores actifs placés à proximité de la zone à renforcer.

Here are some typical applications for active sound reflectors:

• Reinforcement of direct sound, thanks to active sound reflectors placed above the proscenium, or near the stage frame.
• Active stage shell. Improvement of the acoustics on stage thanks to sound reflectors placed around the musicians.
• Increase the reverberation time of a room thanks to a set of active sound reflectors.
• Creation of a room effect in an outdoor show.
• Local reinforcement of the sound level in the room, through the use of one or more active sound reflectors placed near the area to be reinforced.

Active sound reflectors are radically different of sound in that they basically operate a modification of the acoustics of the room, rather than a massive diffusion of sound energy. So in a room equipped with a set of active sound reflectors well settled, the listener has the impression of natural acoustics not amplified.

Passive shutters have been used in rooms for many years. They are in the form of reflective, flat or curved panels, generally suspended above the stage or the proscenium. Their typical dimensions are of the order of one to a few square meters. The idea of active sound reflector is described in the French patent application FR 2 449 318 filed on 11-02-1980 by Philips Gloeilampenfabrieken NV where we speak of a "sound amplifier reflector" comprising a microphone, an amplifier and a loudspeaker whose direct radiation is hardly picked up by the associated microphone, the whole being assembled in a “mechanical unit”. In said patent, we do not speak explicitly of the amplification of early energy, but rather of the use of a certain number of these reflectors for the reinforcement of the reverberation in a room. In addition, since the 1960s, a certain number of commercial realizations of electroacoustic systems of reverberation reinforcement have emerged, which one finds a brief description in: M. Kleiner, P. Svensson, Review of active systems in room acoustics and electroacoustics . Proc. of ACTIVE 95, Newport Beach, CA, USA, 1995, 39-54. These include MCR (Multichannel Reverberation from Philips: described in SH De Koning, The MCR system: multiple-channel amplification of reverberation. Philips Tech. Rev., vol. 41, no. 1, pp 12-23, 1983 / 84), ACS (Acoustic Control System, described in: AJ Berkhout, DD De Vries, P. Vogel, Acoustic control by wave field synthesis . J. Acoust. Soc. Am., Vol 93, no. 5, pp 2764-2778 , 1993; and in European patent application EP 0 335 468 A1 filed by Birch Wood Acoustics Nederland BV), SIAP (System for Improved Acoustic Performance, described in: WCJM Prinssen, M. Holden, System for improved acoustic performance . Proc. IOA, pp 93-101, 1992; and in European patent application EP 0 386 846 A1 filed by Prinssen En Bus Raadgevende Ingenieurs VOF), LARES (Lexicon Acoustic Reverberation Enhancement System, described in: D. Griesinger, Improving room acoustic through time-variant synthetic reverberation. Audio Eng. Soc. 90 ^th Convention, Preprint 3014 (B-2 ), 1991; and in US patent 5,109,419 to Lexicon Inc.), and VRA (Variable Room Acoustic, described in: MA Poletti. On controlling the apparent absorption and volume in assisted reverberation system. Acta Acustica, vol. 78, pp 61-73, 1993; and in US patents US 5,729,613 and US 5,862,233 from Industrial Research Limited). All these systems consist of a number of microphones associated with loudspeakers via a multichannel electronic amplification and filtering chain. These are looped systems since the signals picked up by the microphones are sent to the speakers, whose acoustic radiation is picked up by these same microphones. The microphones are distributed in the room and / or near the stage, and the loudspeakers distributed in the room (and sometimes also on the stage). However, in these systems the sound picked up by a microphone is sent by the electronic chain to a speaker located far from this microphone. In fact, otherwise, the significant energy of the direct acoustic wave propagating from the loudspeaker towards the microphone would impose a very low electronic amplification gain so that the looped system does not become unstable (Larsen effect), thereby compromising the efficiency of the system. Only the Carmen system (CSTB-France) uses the concept of "proximity reaction". In this system, the signal from a microphone is sent ( via an electronic chain) to a speaker located near the microphone, and the energy of the direct acoustic wave speaker → microphone is limited by a principle acoustic decoupling: the directivities and positions of the microphone and the associated loudspeaker are chosen so that the microphone picks up little direct energy from the loudspeaker. We speak here of "proximity reaction" when the distance between the microphone and the associated loudspeaker is small compared to the typical propagation distances in rooms, but all the same greater than the shortest wavelengths considered (a few centimeters at very high frequencies). Although this definition is not explicitly mentioned in the patent application FR 2 449 318, it can be considered that it is implicit in the concept of sound reflection mentioned therein. When the distance between microphone and loudspeaker is less than the shortest wavelengths considered, we speak of "local reaction", as is the case in active impedance control applications such as for example that described in: X Meynial, Active Materials for applications in room acoustics , 3rd ICIM / ECSSM '96, Lyon '96, 968-973. In the case of the active sound reflector, said distance is typically of the order of 20 cm to 2 m.

Figure 1 shows the block diagram of the sound reflector active. The microphone (1) is connected to the input of a preamplifier (2) whose output is connected to a circuit control (3). The control circuit output signal (3) is amplified by a power amplifier (4) charged by a loudspeaker (5). The microphone (1) and the speaker (5) are subject to a screen (6). The distance from the microphone (1) of the speaker (5) is around 20cm at 2m typically. The active sound reflector may include several microphones (1) and several preamplifiers (2), and several speakers (5) and several amplifiers Power 4). In this case, the active sound reflector is characterized by the average distance between all pairs (speaker - microphone) of the reflector; this distance being between 20cm and 2m. The functioning of each active reflector is independent insofar as it does not receive electrical signals from other active reflectors.

The device according to the invention corresponds to the description above and is characterized by the functions performed by the control circuit (3). It can be decoupling electronics between speakers (5) and microphones (1), directional filtering, and reverberator. These functions will be explained below.

The microphones (1) can more generally be sound sensors, but we will keep the name of microphone in the suite. The designation "function of transfer ”indicates in the following a function of the frequency, although the variable is not mentioned for do not weigh down the explanations.

The active sound reflector can only work if you limits the acoustic coupling between the speakers (5) and microphones (1), otherwise the system would become unstable (Larsen effect) even at low gain values of the amplification chain (2,3,4). In the following, we exposes the problem of acoustic coupling for a single speaker and only one microphone, the problem being essentially identical when multiple microphones and / or multiple speakers are used.

There are two ways to limit the acoustic coupling ways: acoustic decoupling, and electronic decoupling. We use one or the other, or a combination of two in the active sound reflector. The final decoupling must be such that the energy of the direct sound speaker (5) → microphone (1) is less than the energy reflected by the room coming from the loudspeaker (5), and received by the microphone (1). It is on this condition that a set of some active sound reflectors will be able to reflect on the audience of sound waves whose amplitude is comparable to that of the direct wave reaching the listeners coming from the stage.

Acoustic decoupling consists of playing on the directivity characteristics of the loudspeaker (5) and / or the microphone (1) and on the symmetry of the device, so that the direct sound picked up by the microphone from the speaker either of low amplitude. A certain number of combinations are possible, such as those illustrated in figure 2 which use either a microphone bidirectional (Figure 2a), i.e. a microphone omnidirectional (Figure 2b), i.e. a cardioid microphone (Figure 2c). The reflector is seen from the front in FIG. 2a, and in section in FIGS. 2b and 2c. In Figures 2a and 2c, the dotted arrow represents the direction of the maximum microphone sensitivity. The idea of decoupling acoustics is presented in patent application FR 2 449 318, which describes the implementation corresponding to the figure 2c. It is important to note that some of these solutions use directional microphones, which allows, if we point the axis of maximum sensitivity of these towards the scene, to privilege the capture of the direct sound in origin of the scene in relation to sound capture reverberated by the room, especially since the reflector is close to the stage. The efficiency of the reflector in terms reinforcement of early energy is increased. The judicious use of several microphones (forming a microphone antenna) can allow a capture even more selective from the source on stage. Likewise, using multiple speakers can allow favor direct propagation between the speakers and listeners versus the reverberant spread by the room.

In the device according to the invention, the electronic decoupling possibly completes the acoustic decoupling to attenuate the influence of the sound reflections coming from obstacles located near the active reflector, diffracted waves on the edge of the screen (6) , or waves reflected by other neighboring active sound reflectors. Electronic decoupling consists in attenuating the electrical signal delivered by the microphone (1) coming from the loudspeaker (5) using an echo canceller filter (7) included in the control circuit (3), and as shown in figure 3. In this filter, the transfer function K = -V _m / V _hp must be estimated (in the absence of feedback: X = G = 0) while the reflector is installed in the room, and the estimate X injected into the echo canceller filter (7). The transfer function K can either be measured beforehand by applying a test signal to the input of the power amplifier (4), or estimated continuously during the operation of the reflector according to adaptive techniques well known in public address or telephony. . The echo canceller (7) cancels the most energetic part of the loudspeaker (5) → microphone (1) response, ie the start of the corresponding impulse response (typically the first milliseconds). The transfer function G in FIG. 3 determines the characteristics of the sound reflected by the active reflector.

The use of several loudspeakers (5) excited by delayed signals and adequately filtered by the control circuit (3) can make it possible to control the directivity diagram (i.e. the sound level emitted as a function of the direction considered) of the loudspeaker network (5), according to the well-known antenna principles. For example, a vertical linear array of loudspeakers separated by a distance d (assumed to be less than the considered half-wavelengths) and indexed from 1 to p from the top, radiates a maximum of energy in the direction  if a delay τ _n = n.sin () .d / c is assigned to the signal sent to the nth loudspeaker, c being the speed of the sound waves, and  = 0 designating the horizontal. This elementary principle easily extends to a two-dimensional network. We can then define the “emission direction” of the network as the angular direction corresponding to the maximum of energy emitted. Thus, the device according to the invention provided with several speakers (5) makes it possible to generate several reflections each characterized by a delay, an attenuation, and a direction of emission.

où R_i est la fonction de transfert en tension du réverbérateur i, Or, il est bien connu en acoustique des salles que les distributions fréquentielles des modules des fonctions H_i sont régies par la loi de Rayleigh si le son direct haut-parleurs(5)→ microphone(1) est de faible énergie par rapport au son réverbéré par la salle (condition qui est assurée par les découplages acoustiques et électroniques). Plus le rayonnement du réseau de haut-parleur (5) pour une direction d'émission donnée (correspondant à un des réverbérateurs) se distingue du rayonnement correspondant aux autres directions d'émission (correspondant aux autres réverbérateurs), moins les fonctions H_i sont corrélées entre elles. Par ailleurs, les distributions des modules des fonctions de transfert R_i sont également régies par la loi de Rayleigh puisqu'elles traduisent aussi un phénomène de réverbération. Alors, par application de la loi des grands nombres, la variance de |T| diminue et sa distribution tend vers une loi de Rayleigh à mesure que le nombre n de réverbérateurs (9) augmente, et ce d'autant plus que les fonctions H_i sont décorrélées entre elles. Ceci présente un grand intérêt en pratique, puisque le gain de la chaíne d'amplification (2,3,4) peut être d'autant plus fort que la distribution de |T| approche la loi de Rayleigh. A la limite. le gain sera le même que s'il n'y avait pas de réverbérateur, mais l'utilisation de ceux-ci permettra de prolonger efficacement la réverbération de la salle. Un tel réflecteur sonore actif permet donc à la fois de générer des réflexions précoces de forte amplitude (en tirant parti de la directivité du réseau de haut-parleurs), et d'augmenter la durée de réverbération d'une salle. On peut encore faire varier lentement le filtrage effectué par la matrice directionnelle (10) de sorte à faire évoluer dans le temps la direction d'émission assignée à chaque réverbérateur, ce qui permettra d'améliorer la stabilité du système et de réduire la coloration liée au rebouclage acoustique des haut-parleurs (5) vers les microphones (1). Un mise en oeuvre efficace de ce principe consiste à effectuer une rotation lente des affectations des sorties des réverbérateurs vers les directions d'émission. Le réseau de haut-parleurs (5) peut prendre diverses formes, comme par exemple le réseau plan ou le réseau linéaire. Une combinaison particulièrement avantageuse est obtenue avec un réseau linéaire vertical de haut-parleurs (5) et un microphone bidirectionnel (1) dont la membrane se situe dans un plan contenant les haut-parleurs (5), assurant ainsi un excellent découplage acoustique selon le principe de la figure 2a. Une version simplifiée du dispositif consiste à supprimer la matrice directionnelle (10), et affecter la sortie de chaque réverbérateur (9) directement à un ou plusieurs haut-parleurs du réseau (5). Outre l'intérêt qui vient d'être discuté, le fait d'utiliser plusieurs haut-parleurs (5) permet de répartir l'énergie sonore émise par le réflecteur sur une plus grande surface, et par là même d'éviter qu'un auditeur ne soit gêné par une trop forte densité d'énergie sonore en provenance d'un seul haut-parleur (5). Si l'on retire les réverbérateurs (9), on conserve ce dernier avantage, mais on perd le bénéfice de ceux-ci pour l'augmentation de la durée de réverbération de la salle. Le réflecteur aura cependant toujours tendance à augmenter la réverbération par simple rebouclage acoustique des haut-parleurs (5) vers le microphone (1) via la salle. Dans la suite, le cas où il n'y a pas de réverbérateur est considéré comme un cas particuliers de réverbérateur de fonction de transfert unitaire R=1.FIG. 4 represents an embodiment of the invention consisting of a device using a microphone (1), a network of speakers (5), and a number n of reverberators (9) whose output signals are assigned to as many emission directions, thanks to a “directional matrix” (10). The electronic decoupling corresponding to the transmission direction i is ensured by the echo canceller filter (7) of transfer function X _i . By noting V _m the voltage delivered by the preamplifier (2) of the microphone (1), and V _Ri the output voltage of the reverberator i, then the transfer function H _i = V _m / V _Ri represents the emission of the reflector in direction i and the acoustic propagation in the room up to the microphone (1). Note that this function takes into account the echo canceller filter X _i : H _i = K _i + X _i . The gain in open loop of the {reflector + room} system is therefore written

where R _i is the voltage transfer function of the reverberator i, However, it is well known in room acoustics that the frequency distributions of the modules of the functions H _i are governed by Rayleigh's law if the direct sound speakers (5 ) → microphone (1) is of low energy compared to the sound reverberated by the room (condition which is ensured by acoustic and electronic decoupling). The more the radiation from the loudspeaker network (5) for a given direction of emission (corresponding to one of the reverberators) differs from the radiation corresponding to the other directions of emission (corresponding to the other reverberators), the less the functions H _i are correlated with each other. Furthermore, the distributions of the modules of transfer functions R _i are also governed by Rayleigh's law since they also reflect a phenomenon of reverberation. Then, by applying the law of large numbers, the variance of | T | decreases and its distribution tends towards a Rayleigh law as the number n of reverberators (9) increases, and this all the more that the functions H _i are uncorrelated with each other. This is of great interest in practice, since the gain of the amplification chain (2,3,4) can be all the stronger as the distribution of | T | approach Rayleigh's law. At the limit. the gain will be the same as if there were no reverberator, but the use of these will effectively prolong the reverberation of the room. Such an active sound reflector therefore makes it possible both to generate early reflections of high amplitude (by taking advantage of the directivity of the network of loudspeakers), and to increase the reverberation time of a room. The filtering carried out by the directional matrix (10) can still be varied slowly so as to change over time the direction of emission assigned to each reverberator, which will make it possible to improve the stability of the system and reduce the coloration linked acoustic feedback from the speakers (5) to the microphones (1). An effective implementation of this principle consists in carrying out a slow rotation of the assignments of the outputs of the reverberators towards the directions of emission. The network of loudspeakers (5) can take various forms, such as for example the flat network or the linear network. A particularly advantageous combination is obtained with a vertical linear array of speakers (5) and a bidirectional microphone (1) whose membrane is located in a plane containing the speakers (5), thus ensuring excellent acoustic decoupling according to the principle of Figure 2a. A simplified version of the device consists in removing the directional matrix (10), and assigning the output of each reverberator (9) directly to one or more speakers of the network (5). In addition to the interest which has just been discussed, the fact of using several loudspeakers (5) makes it possible to distribute the sound energy emitted by the reflector over a larger surface, and by this same to avoid that a listener is not bothered by too high a density of sound energy coming from a single speaker (5). If we remove the reverberators (9), we keep this last advantage, but we lose the benefit of them for the increase in the reverberation time of the room. The reflector will however always tend to increase the reverberation by simple acoustic looping from the speakers (5) to the microphone (1) via the room. In the following, the case where there is no reverberator is considered as a particular case of reverberator of unit transfer function R = 1.

FIG. 5 illustrates another embodiment of the device according to the invention. Similar to the above, the use of a network of microphones (1) and a directional matrix (10) placed between the microphone preamplifiers (2) and the reverberators (9) makes it possible to pick up the incident waves according to several directivity diagrams (we will speak of "sensor channels"). each of these sensor channels being assigned to a separate reverberator (9). The echo cancellation filter has been omitted for clarity. The outputs of the reverberators (9) are sent to a summator (10), the output of which excites the loudspeaker (5) via the power amplifier (4). The open loop transfer function of the system is of the same form as that given in the previous paragraph, and therefore has the same advantages in terms of reinforcement of the reverberation. As in the previous case, a simplified implementation of the device consists in eliminating the directional matrix (10) and in assigning the output of each microphone (1) directly to one or more reverberators (9). The directivity of the capture is then determined by the directivity of each microphone (1).

Finally, it is of course possible to combine as shown Figure 6 the systems described in the last two paragraphs: we then use a network of microphones (1) connected via their preamplifiers (2) to a matrix directional capture (10a), and a network of loudspeakers (5) energized via their power amplifiers (4) by the outputs of a directional emission matrix (10b); the reverberators (9) being placed between the outputs of the directional capture matrix (10a) and the inputs of the directional emission matrix (10b). With this last embodiment of the following device the invention, we can assign to each wave received according to a given channel any number of waves possibly delayed and attenuated relative to each other at each radiation path.

Figure 7 illustrates in a simple way this type of possibility, with a reflector based on 2 microphones cardioids (1) pointing in the directions +  and -, and a array of loudspeakers (5) separated by a distance d, reflecting waves in a “pseudo-specular” way thanks to to a directional emission matrix (10b) consisting of summers (8) and delays (11) τ = d / c.sin (): at the incident direction +  corresponds to the reflected direction - (and vice versa), but only two directions are considered. In this example, there is no directional matrix of capture because the directivities of two capture channels simply result from the directivity of the microphones (1). You can optionally introduce reverberators between the outputs of the microphone preamplifiers (2) and the inputs of the directional emission matrix (10b).

In Figures 5 to 7, the echo canceller filter has been omitted to clarify more. It is clear that the structure and response of it tend to become more complex to measure that the number of microphones or speakers is increased and that the number of reflections issued for each incident wave increases. However, it should be clarified that, apart from any consideration linked to the directivity of a microphone or speakers, multiplying the number speakers (not accompanied by a multiplication of the total power emitted because the gain in medium open loop <T> must remain constant) tends to decrease the energy of the direct sound picked up by the microphone by where all the speakers come from if the speakers added are located farther from the microphone than those existing.

In addition to the electronic decoupling function and the operations linked to the use of several microphones and / or speakers, the control circuit (3) provides various filtering functions, such as gain control, equalization (frequency correction). , the compression of the signal dynamics to limit the maximum amplitude of the signals sent to the loudspeaker (5) via the power amplifier (4), or even a filtering varying in time (slow modulation of the phase or of the signal delay, or more complex modulation) which possibly makes it possible to further increase the gain threshold of the control circuit corresponding to instability (Larsen effect). Most of the filtering operations of the control circuit (3) are carried out by one or more digital signal processors (DSP). These filtering functions, which have not been explicitly mentioned in the preceding paragraphs and the accompanying figures for the sake of clarity, can be considered to be part of the reverberators (9).
The user adjusts the behavior of the active sound reflector by modifying (using a remote control) the filtering parameters of the control circuit (3).

All of the electronics {preamplifier (2), circuit control (3). power amplifier (4)} can be physically integrated in the reflector, for example on the back of the screen (6), or remote. The preamplifiers (2) can possibly supplying power for microphones (1), especially if these are of the electrostatic. They can possibly be part of the same mechanical unit as the control circuit (3), everything like power amplifiers (4). Depending on the type of directionality desired for the speakers (5), these can be mounted in a closed or bass-reflex enclosure, or simply placed in a screen, or any other type of acoustic load. The microphones (1) are either placed on the screen, or deported (as is the case in Figure 2c) using a stick for example. The screen (6) can be of various dimensions and shapes (rectangular, elliptical, or other), provided that decoupling is sufficient. If the speakers are mounted in a pregnant, the screen may be reduced to the front of the enclosure. It is not necessarily plan.

Although the operation of each active reflector is autonomous, several active sound reflectors will be often associated, as in the first four examples of use mentioned above. In this case control of control parameters is common to the whole associated sound reflectors, and transmitted by a network shared by these sound reflectors.

The shape of the sound reflectors may be designed with a view to of the assembly of several of them, for example for form a large active reflector above the stage frame.

Claims (13)

Active device reflecting sound waves and intended for improving listening conditions in concert halls, theaters and amphitheatres, recording studios, etc., comprising one or more sound sensors (1) and their preamplifiers (2 ), and one or more loudspeakers (5) and their power amplifiers (4), characterized in that the average distance between sensor and loudspeaker for all the pairs (sensor - loudspeaker) is between twenty centimeters and two meters, and in that the inputs of the amplifiers (4) are connected to the outputs of the preamplifiers (2) via an electronic control circuit (3) ensuring in particular at least one of the following functions: echo cancellation reducing the acoustic coupling between the speakers (5) and the sound sensors (1), control of the directivity of all the sound sensors (1), control of the directivity of all the speakers (5), reverberator filter. Device according to claim 1, comprising a network of several loudspeakers (5) and a control circuit (3) provided with one input and several outputs connected to the loudspeakers (5) via several amplifiers power (4), characterized in that said control circuit (3) makes it possible, from a wave picked up by the sensor (s) (1), to emit any number of waves according to several directivity diagrams of the high network -speakers (5), and possibly delayed and attenuated with respect to each other. Device according to claim 1, comprising a network of several sound sensors (1) and a control circuit (3) provided with an output and several inputs connected to the sensors (1) via several preamplifiers (2), characterized in that said control circuit (3) allows the capture of waves according to several directivity diagrams of the network of sensors each corresponding to a "sensor channel", and emitting any number of waves possibly delayed and attenuated relative to each other to the others for each wave received by each sensor channel. Device according to claim 1, comprising: a network of several sound sensors (1), a network of several speakers (5), and a control circuit (3) provided with several inputs connected to the sensors (1) via several preamplifiers (2), and several outputs connected to the speakers (5) via several power amplifiers ( 4), characterized in that said control circuit (3) allows the capture of waves according to several directivity diagrams of the network of sensors (1) each corresponding to a "sensor channel", the emission of waves according to several directivity diagrams of the high network -speakers (5) each corresponding to a "radiation channel", and comprises a matrix of reverberators whose inputs come from the sensor channels and the outputs are assigned to the radiation channels. Device according to claim 2, characterized in that it uses several loudspeakers (5) aligned vertically and mounted in a column type enclosure, a sound sensor (1) consisting of a bidirectional microphone fixed to the column speaker and placed so that its membrane describes a plane containing the loudspeakers (5). Device according to claim 2, characterized in that the control circuit (3) contains several reverberant filters excited by the same signal from the preamplifier (2), is provided with as many outputs as there are reverberant filters, each of these outputs being connected to a power amplifier (4) charged by one or more speakers (5), the assignment between the outputs of the reverberators and the outputs of the control circuit (3) may possibly vary slowly during the time. Device according to claim 5, characterized in that the control circuit (3) contains several reverberant filters excited by the same signal from the preamplifier (2), and is provided with as many outputs as there are reverberant filters, each of these outputs being connected to a power amplifier (4) charged by one or more speakers (5), the assignment between the outputs of the reverberators and the outputs of the control circuit (3) may possibly vary slowly during time. Device according to claim 2, characterized in that the control circuit (3) contains several delays excited by the same signal from the preamplifier (2), is provided with as many outputs as there are delays, each of these outputs being connected to a power amplifier (4) loaded by one or more loudspeakers (5), the assignment between the outputs of the delays and the outputs of the control circuit (3) possibly being able to vary slowly over time. Device according to Claim 4, characterized in that the characteristics of the sensor channels and of the radiation channels, as well as the properties of the matrix of reverberators may possibly vary slowly over time. Device according to claim 9, characterized in that the sensor channels are each constituted only by the signal from a sound sensor or a group of specific sound sensors (1). Device according to claim 9, characterized in that the radiation paths are each constituted only by a specific loudspeaker or group of loudspeakers (1). Device according to Claim 10, characterized in that the radiation paths are each constituted only by a specific loudspeaker or group of loudspeakers (1). Device according to Claim 10, characterized in that it uses a plane array of speakers (5) and two sound sensors (1) consisting of two cardioid microphones pointing in two directions symmetrical with respect to the plane of the speakers (5).

Images