EP3586328B1 - Augmented acoustic music instrument with feedforward and feedback actuators . - Google Patents

Description

The present invention relates to a processing of acoustic data picked up on a musical instrument having an acoustically radiating structure. More specifically, provision is made to supply one or more actuators of the radiating structure of the instrument with a signal generated from the acoustic data captured and processed, with a view to enriching the vibratory properties and in particular the sound resulting from the instrument with desired sound effects (echo, reverb, distortion, equalization, etc.).

For example, string musical instruments have a radiating structure (soundboard and possibly sound box) coupled to a bridge carrying the strings. It is then proposed in the context of the present invention to cause the radiating structure to resonate with a particular effect, in addition to the playing of the musician. For example in the case of an echo (or "delay" in English), the musician plays a note which the radiating structure amplifies and diffuses, but in addition, one or more actuators acting on the radiating structure then apply a vibration to the structure to replay this note at regular time intervals with a decrease in amplitude to simulate the echo effect.

This approach differs from the case of effects applied classically by playing typically on an electric guitar connected to an amplifier via a cable (or "jack"). With reference to the figure 1 illustrating this case, one or more MIC pickup(s) mounted on the guitar GUI pick up the vibration signal from the strings and this latter feeds an EF device applying a chosen transformation of the signal (echo, reverberation, distortion, equalization, "phaser" type phase change or slower "flanger" type change, a slight frequency change with "chorus" type or more marked "octaver" type mixing, "tremolo" type amplitude modulation , a change in sound amplitude: dynamically (type "sustain" or "compression" or not, or others) This EF device (commonly called "effect pedal" or "effects pedalboard") is conventionally connected to an AMP amplifier which electronically amplifies and radiates the acoustic signal transformed by the EF effects pedalboard.

WO 2016/209143 A1 describes a method and process for determining the parameters of a control algorithm to synthesize and generate a variety of sounds and vibrations not normally available on a stringed instrument. WO 97/12359 A1 describes a hybrid active vibration attenuation technique for noise canceling headphones. WO 2016/209143 A1 And WO 97/12359 A1 do not, however, describe the aforementioned transformations of the signal.

In the case of the approach within the meaning of the present invention, the radiating structure of the instrument, itself (typically the CAI resonance box of a guitar for example), is used as a "diffuser" or "high- speaker” of the sound signal transformed by a device DEV of the “effects pedalboard” type.

Especially in the example of the picture 2 , one or more MIC sensors are mounted on the sound box of the guitar (for example at the soundhole). This (these) sensor(s) capture(s) the acoustic vibrations of the radiating structure. The digital signal corresponding to this acoustic signal is transmitted to the input E of a device DEV applying the desired effect(s) and controlling, via its output S, actuators ACT applied against the resonance box CAI to make the box vibrate according to the effects chosen by the user of the device DEV.

• Intégrer des capteurs et actionneurs à ces instruments
• Appliquer des traitements sur les signaux captés
• et renvoyer ces signaux captés aux actionneurs.

Generally, in the context of picture 2 , the electronic transformation of the sound radiated by a stringed instrument has so far consisted of:

• Integrate sensors and actuators into these instruments
• Apply processing to the captured signals
• and returning these sensed signals to the actuators.

The sound radiated by the instrument is thus the sum of the acoustic sound played by the musician and its transformations by the DEV device (without the need to pass the captured signal through an amplification chain, as done conventionally and illustrated on the figure 1 ).

The transformations thus applied are generally digital audio effects (reverberation (or "reverb"), chorus, distortion, equalization) injected in "feedforward", that is to say that the processing does not take into account the feedback emitted by the actuators on the sensors.

Transformations applied with these techniques do not achieve the desired effects.

Digital audio effects induce instabilities (Larsen effect). This means an undesired frequency superimposed on the desired signal.

The sound radiated has a poor quality, for example compared to another instrument or to that obtained by a conventional amplification chain of the type illustrated in the figure 1 .

These two defects come from the fact that the characteristics of the radiating structure and/or of its coupling with the excitation by the strings are not taken into account. Indeed, the vibratory characteristics of the radiating structure transform the signals emitted by the actuators in a uneven across frequencies. This is due in particular to the regions of the body where the resonance modes induce amplitude modifications from one frequency to another. This uneven characteristic is imposed by the manufacturer of the instrument and is indicative of the quality of the instrument when played by plucking the strings. On the other hand, when the excitation is done by the actuators, this induces an uneven sound quality depending on the notes played. Moreover, the strong coupling between the strings and the body at certain frequencies induces a strong feedback on the sensors after emission by the actuators. This feedback changes the frequencies and damping of the body resonances. The fact of not taking this feedback into account is thus a source of error and instability of the targeted sounds.

The present invention as defined by the claims improves the situation.

• la figure 1 illustre le montage classique d'un instrument connecté à un pédalier d'effets, lui-même connecté à un amplificateur,
• la figure 2 illustre un montage au sens de l'invention d'un capteur et d'un ou plusieurs actionneurs sur un instrument connecté à un dispositif gérant les actionneurs notamment en fonction d'une consigne d'un utilisateur du dispositif,
• la figure 3 illustre la transformation du timbre d'un instrument, ici par simple contrôle de type feedforward modifiant la pression acoustique rayonnée p (chemin primaire depuis l'excitation de la corde), et montre en particulier que le chemin secondaire (de l'actionneur au capteur) peut induire une instabilité, en l'absence de contrôle de la rétroaction ;
• la figure 4 illustre un ajustement d'un contrôle de rétroaction type « feedback » (FB) suite à la mesure de la fonction de transfert entre capteur et actionneur en boucle ouverte ;
• la figure 5 illustre un ajustement de contrôle de type feedforward (FF), en fonction de l'effet choisi par le musicien ;
• la figure 6 illustre un ajustement en parallèle du contrôle de rétroaction, actualisé pour prendre en compte les nouvelles valeurs du contrôle feedforward imposées par la consigne de l'effet choisi par le musicien ;
• la figure 7 illustre un ordinogramme représentant les étapes d'un exemple de procédé au sens de la présente invention ;
• la figure 8 illustre un exemple de dispositif pour la mise en oeuvre de l'invention ;
• la figure 9 illustre un exemple de réalisation avantageux d'un équipement pour une guitare, connecté à un dispositif au sens de l'invention ;
• les figures 10A, 10B et 10C illustrent des traitements opérés dans un exemple de réalisation pour obtenir des paramètres déterminés à partir de la fonction de transfert H1 précitée, en vue du contrôle feedforward.

Other advantages and characteristics of the invention will appear on reading the detailed description below of exemplary embodiments of the invention, and on examining the appended drawings in which:

• there figure 1 illustrates the classic assembly of an instrument connected to an effects pedalboard, itself connected to an amplifier,
• there figure 2 illustrates an assembly within the meaning of the invention of a sensor and one or more actuators on an instrument connected to a device managing the actuators in particular according to an instruction from a user of the device,
• there picture 3 illustrates the transformation of the timbre of an instrument, here by simple feedforward type control modifying the radiated acoustic pressure p (primary path from the excitation of the string), and shows in particular that the secondary path (from the actuator to the sensor ) can induce instability, in the absence of feedback control;
• there figure 4 illustrates an adjustment of a "feedback" (FB) type feedback control following the measurement of the transfer function between sensor and actuator in open loop;
• there figure 5 illustrates a feedforward (FF) control adjustment, depending on the effect chosen by the musician;
• there figure 6 illustrates a parallel adjustment of the feedback control, updated to take into account the new values of the feedforward control imposed by the setpoint of the effect chosen by the musician;
• there figure 7 illustrates a flowchart representing the steps of an exemplary method within the meaning of the present invention;
• there figure 8 illustrates an example of a device for implementing the invention;
• there figure 9 illustrates an advantageous embodiment of equipment for a guitar, connected to a device within the meaning of the invention;
• THE figures 10A, 10B and 10C illustrate processing carried out in an exemplary embodiment to obtain parameters determined from the aforementioned transfer function H1, with a view to feedforward control.

• d'un capteur piézoélectrique CAP sous le sillet (partie sous le chevalet portant les cordes),
• d'un ou plusieurs (par exemple deux) actionneurs ACT électrodynamiques montés ici en parallèle sur chaque côté du chevalet, et
• d'un dispositif DIS (connecté par son entrée E au capteur, et sa sortie S aux actionneurs).

As illustrated on the figure 9 , an acoustic guitar equipped with a device within the meaning of the invention is provided with:

• a CAP piezoelectric sensor under the saddle (part under the bridge carrying the strings),
• one or more (for example two) electrodynamic ACT actuators mounted here in parallel on each side of the bridge, and
• a DIS device (connected by its input E to the sensor, and its output S to the actuators).

• un préamplificateur PRA pour le capteur (via l'entrée E du dispositif),
• un convertisseur analogique-numérique rapide CAN,
• un microcontrôleur CTL,
• un convertisseur numérique-analogique rapide CNA et un amplificateur de puissance AP excitant les actionneurs ACT (via la sortie S du dispositif).

With reference to the figure 8 showing in detail the DIS device in an exemplary embodiment, the device comprises:

• a PRA preamplifier for the sensor (via the E input of the device),
• a fast CAN analog-to-digital converter,
• a CTL microcontroller,
• a fast digital-analog converter CNA and a power amplifier AP exciting the actuators ACT (via the output S of the device).

The physical latency of processing does not exceed a few microseconds.

• une mémoire MEM stockant des données d'instructions d'un programme informatique au sens de l'invention (et éventuellement d'autres données non permanentes, de calcul), et
• un processeur PROC lisant le contenu de la mémoire MEM pour exécuter le programme informatique, en mettant ainsi en oeuvre des algorithmes de traitements audionumériques effectués par échantillon, ces algorithmes étant informés par une estimation des propriétés de la structure rayonnante, obtenues comme détaillé ci-après.

Thus, the device DIS operates practically in real time (at very low latency such as for example a few microseconds between the input E and the output S). The DIS device comprises a microcontroller or more generally a CTL processing circuit typically including:

• a memory MEM storing instruction data of a computer program within the meaning of the invention (and possibly other non-permanent, calculation data), and
• a processor PROC reading the contents of the memory MEM to execute the computer program, thus implementing digital audio processing algorithms performed per sample, these algorithms being informed by an estimate of the properties of the radiating structure, obtained as detailed below .

• une fonction de transfert H1 entre le capteur CAP et les actionneurs ACT est estimée initialement en boucle ouverte comme illustré sur la figure 4,
• un traitement acoustique (par exemple un effet ou une combinaison d'effets) est présélectionné par un utilisateur via une interface homme/machine (IHM) que comporte le dispositif DIS,
• le contrôleur CTL ajuste éventuellement la fonction de transfert estimée, en fonction de l'effet programmé,
• lorsque l'utilisateur joue de l'instrument, l'effet programmé est appliqué pour mettre en oeuvre les actionneurs, en mode feedforward (flèche F1 de la figure 3),
• ensuite, la vibration que fait opérer les actionneurs sur l'instrument et notamment sur les cordes est prise en compte (flèche F2 de la figure 3) en tenant compte de la fonction de transfert ajustée, et en contrôlant en particulier le signal que capte le capteur CAP (par exemple en prévoyant un contrôle au niveau de la pré-amplification PRA par le processeur PROC comme illustré sur la figure 8),
• le son ou la vibration capté par le capteur CAP est ainsi ajusté et analysé en mode feedback pour appliquer l'effet désiré (CTL FF) avec la prise en compte de l'activation des actionneurs sur la vibration des cordes et plus généralement de la structure rayonnante, cette vibration venant s'ajouter au jeu naturel du musicien et à l'effet acoustique désiré.

The present invention proposes feedback/feedforward (FB/FF) type processing, in which:

• a transfer function H1 between the sensor CAP and the actuators ACT is initially estimated in open loop as illustrated in the figure 4 ,
• an acoustic treatment (for example an effect or a combination of effects) is preselected by a user via a man/machine interface (HMI) that comprises the device DIS,
• the CTL controller possibly adjusts the estimated transfer function, depending on the programmed effect,
• when the user plays the instrument, the programmed effect is applied to activate the actuators, in feedforward mode (arrow F1 on the picture 3 ),
• then, the vibration caused by the actuators on the instrument and in particular on the strings is taken into account (arrow F2 of the picture 3 ) taking into account the adjusted transfer function, and in particular controlling the signal picked up by the CAP sensor (for example by providing control at the level of the pre-amplification PRA by the processor PROC as illustrated in figure figure 8 ),
• the sound or vibration picked up by the CAP sensor is thus adjusted and analyzed in feedback mode to apply the desired effect (CTL FF) taking into account the activation of the actuators on the vibration of the strings and more generally of the structure radiant, this vibration adding to the natural playing of the musician and the desired acoustic effect.

It is also possible to estimate in real time the vibroacoustic transfer function H2 between the actuators and one or more acoustic microphones positioned at any point in space to measure the pressure p (close to the ears of the musician, the listener, or even a sound recording, for example from a smartphone integrating the computer means of a device within the meaning of the invention). Thus, for example, the aforementioned preselection of a particular processing for a sound effect chosen by the user can be carried out statically by an application on a smartphone, typically via a wireless connection (bluetooth for example), or dynamically directly on the instrument (for example with potentiometers like on electric guitars but to directly adjust effects and not volumes).

Thus, the acoustic pressure p presented on the figures 3 to 6 can be measured by a microphone (that of the smartphone used as a user interface for example). This measurement can then be used (in addition to the transfer function H1) in the determination of the gains of the feedforward, or even for the determination of the gains coupled in feedback/feedforward, for an enrichment of the final rendering, to the ear of the musician.

It should be noted that the feedback control mode is not represented on the picture 3 simply illustrating "acoustic paths", but rather on the figure 6 illustrating an implementation of the invention.

In a particular embodiment illustrated on the figure 4 , the transfer function H1 between the sensor and the actuators is measured initially with muted strings (without the musician playing on the strings). This transfer function presents a series of peaks in the frequency space, as well as a average amplitude per frequency band (nine bands for example). It is thus a measurement of the transfer function between actuators and sensors, in open loop, the vibratory properties of the radiating structure then being estimated (frequencies, resonance quality factors, amplitudes at the sensors and at the actuators, and /or other properties). Then, from these measurements, it is deduced from the characteristics of vibration to the CAP sensor which make it possible to refine the control of the feedback to be applied (thanks to methods of automatic estimation of parameters described later). The feedback controller is then programmed from these measurements and estimates. As will be seen below, it is also reprogrammed automatically for each new feedforward processing.

Furthermore, with reference now to the figure 5 , when the user begins to choose the modifications of the acoustic sound that he wishes (effects mentioned above), the feedforward type gains are adjusted. The values of these gains update the transfer function as explained above (since the characteristics of the sound at the pickup will be influenced by the type of effect chosen, such as for example an echo causing the structure to vibrate after the musician's attack ), which also updates the gains of the controller by feedback. One thus obtains a perfect taking into account of the modifications chosen by the musician for an optimal restitution of the instrument (by taking into account the influence of these modifications on the feedback which is intrinsic to the instrument).

• pour plusieurs bandes de fréquences (une dizaine typiquement), et
• en fonction de plusieurs niveaux d'amplitude sonore (caractérisant par exemple le niveau d'excitation à l'attaque du musicien).

If the guitarist chooses for example to increase the sound level by 6dB (doubled sound level), the device measures the modifications of the transfer function H1 in feedforward open loop with the signal to the sensor increased by 6dB. It thus estimates the new frequency amplitude values, as well as their deviation from the initial values. It will thus be understood that the transfer function is estimated preferentially:

• for several frequency bands (about ten typically), and
• according to several levels of sound amplitude (characterizing for example the level of excitation at the attack of the musician).

The controller adjusts the feedback type gains (linked to the 6dB increase of each control gain for example) to obtain stable control. Indeed, if this feedback was not taken into account, the control would generally be unstable. If the musician further changes his sound level by transforming the feedforward gain, the feedback gain is recalculated and applied to the system (device and actuators/sensor).

It will thus be understood that the transfer function is estimated dynamically, in particular as a function of the effect or the combination of effects chosen by the user.

If the musician wants his instrument to have the same timbre as another instrument, such as a guitar of better quality which has been previously analyzed, the amplitudes by bands of this better guitar are targeted by the feedforward gains, these gains updating by elsewhere the characteristics to the sensor. The frequencies and dampings of the best guitar are then targeted by the feedback type controller on the device integrating these gains, by placing the pole of the closed loop system for example. Without the feedback/feedforward combination, the frequencies and dampings are accessible but not the amplitudes per band and instabilities can be generated.

In this case in particular, it may be useful to estimate the aforementioned second transfer function H2 in order to refine the parameters for calculating the feedback (vibratory but also sound), then using at least one microphone to pick up the acoustic pressure p in the air close to the radiating structure of the instrument (for example, simply by the microphone of a nearby smartphone operating the processing of the invention). Thus, the instrument can "sound to the ear" of the user as a chosen target instrument.

$$\mathrm{y}\left(\mathrm{t}\right)=\mathrm{Cx}\left(\mathrm{t}\right)$$

$$\mathrm{y}\left(\mathrm{t}\right)=-\mathrm{Kx}\left(\mathrm{t}\right)$$

où x(t) est le vecteur d'état du système (ensemble des déplacements et vitesses modales par exemple), u(t), y(t) et w(t) étant respectivement la commande, la mesure et la perturbation, A est la matrice caractérisant la structure rayonnante, B celle des actionneurs, C celle du capteur, G celle de la perturbation et K le vecteur gain du contrôleur.By way of purely illustrative and non-limiting example, the instrument/sensor/actuators system including the control can be formalized, in a first classic approach, as follows: $$\mathrm{dx}/\mathrm{dt}=\mathrm{ax}\left(\mathrm{you}\right)+\mathrm{Drank}\left(\mathrm{you}\right)+\mathrm{GW}\left(\mathrm{you}\right)$$

$$\mathrm{there}\left(\mathrm{you}\right)=\mathrm{Cx}\left(\mathrm{you}\right)$$

$$\mathrm{there}\left(\mathrm{you}\right)=-\mathrm{Kx}\left(\mathrm{you}\right)$$

where x(t) is the state vector of the system (set of displacements and modal velocities for example), u(t), y(t) and w(t) being respectively the command, the measurement and the disturbance, A is the matrix characterizing the radiating structure, B that of the actuators, C that of the sensor, G that of the disturbance and K the gain vector of the controller.

This system depends on each radiating structure, the position and quantity of sensors and actuators, and the disturbance.

In a particular embodiment, the capture is carried out using a single piezoelectric sensor (ceramic PZT or PVDF or even MFC for example) under the saddle of the bridge of a guitar or at the interface between the strings and the bridge of a violin. Another embodiment may provide multiple separate pickups on the bridge, one at the interface with each string.

The actuation is such that it produces radiated sound of the quality of a good loudspeaker while allowing the vibration characteristics of the body to be measured. For this, the position and the quantity of actuators can be determined by optimization on a numerical simulation by multi-physics finite elements for example. In a particular embodiment illustrated in the figure 9 , actuation is at the bridge, using two ACT inertial electrodynamic actuators mounted in parallel on either side of the bridge with controllable phase shift or mounted to accommodate a stereo signal.

In the expressions above, the parameters A, B, C and G are estimated for example from numerical calculation on the simulation of the complete electromechanical system with the finite element method. Another approach consists in estimating them experimentally, from the transfer function in open loop between sensor(s) and actuator(s) for A, B and C and an admittance measurement at the easel with an impact hammer or “ vibrating pot” and accelerometer for G. The estimation is then done for example with the Rational Fractional Polynomial (RFP) method. On the other hand, x(t) not being directly accessible (since the measurement gives only y(t)), it is estimated at any time, for example using state observers, like the Luenberger observer.

$$\mathrm{y}/\mathrm{w}=\mathrm{C}\left(\mathrm{sId}-\left(\mathrm{A}-\mathrm{BK}\right)\right){\mathrm{G}}^{-1}\mathrm{pour}\mathrm{le}\mathrm{système}\mathrm{contrôlé}$$

où la fonction de transfert y/w dépend de la variable s, c'est-à-dire que la fonction de transfert y/w peut s'écrire y(s)/w(s) et où Id représente la fonction identité.A transfer function y/w of the system can then be written: $$\mathrm{there}/\mathrm{w}=\mathrm{VS}\left(\mathrm{id}-\mathrm{HAS}\right){\mathrm{G}}^{-1}\mathrm{For}\mathrm{THE}\mathrm{system}\mathrm{alone}$$

$$\mathrm{there}/\mathrm{w}=\mathrm{VS}\left(\mathrm{id}-\left(\mathrm{HAS}-\mathrm{BK}\right)\right){\mathrm{G}}^{-1}\mathrm{For}\mathrm{THE}\mathrm{system}\mathrm{control}$$

where the transfer function y/w depends on the variable s, ie the transfer function y/w can be written y(s)/w(s) and where Id represents the identity function.

The controlled vibration of the radiating structure thus has the dynamics of (A - BK) and no longer that of A alone. The K vector is calculated to reach a certain vibrational target, such as resonance frequencies and dampings. One could for example use pole placement algorithms of (A - BK).

In a second approach presented above with reference to figures 3 to 5 , the proposed controller introduces, in addition in the command, the characteristics of the vibration taken into account at the sensor (allowing to inject a gain in feedforward transforming the radiated acoustic pressure p but generating feedback). In this case, in addition to the estimation of A, B, C and G, an average per frequency band of the transfer function H1 (and potentially of the transfer function H2 illustrated in the drawings) is carried out. For example, nine bands (Hz) can be chosen: [20, 100]; [100, 200]; [200, 400]; [400, 800]; [800, 1600]; [1600, 3200]; [3200, 6400]; [6400, 12800]; [12800, 20000]. The modification of each of these bands thus constitutes the target of the feedforward command. Once this command has been determined, the vector C is calculated.

We have illustrated on the figures 10A, 10B and 10C an example of obtaining the parameters A, B, C, K involved in the above equations. With reference to the figure 10A , the spectrum (amplitudes/frequencies) of the transfer function H1 between the sensor(s) and the actuator(s) is measured. With reference to the figure 10B , the frequency detection of the isolated amplitude peaks of the transfer function H1 makes it possible to obtain the parameters A, B and C. With reference to the figure 10C , the calculation of the average amplitude per frequency band of the transfer function H1 is also carried out in order to obtain the parameter K, also as a function of the previously estimated parameters A, B and C. It is then possible to obtain the gains K of the feedback controller, and also by frequency band those of the feedforward controller. Overall, we thus obtain all the frequencies, dampings and modal gains, with the amplitudes per band.

In what follows, the feedback command is calculated differently compared to the aforementioned first approach, called “classic” (in the sense that it could appear immediately).

In this second approach, equations (1) and (2) remain unchanged but equation (3) becomes: $$\mathrm{a}\left(\mathrm{you}\right)=-\mathrm{Kx}\left(\mathrm{you}\right)+\mathrm{Cx}\left(\mathrm{you}\right)$$

The transfer function y/w of the system is written for the controlled system: $$\mathrm{there}/\mathrm{w}=\mathrm{VS}\left(\mathrm{id}-\left(\mathrm{HAS}+\mathrm{BC}-\mathrm{BK}\right)\right){\mathrm{G}}^{-1}$$

• assurer la stabilité pour toutes les modifications apportées au vecteur C,
• atteindre une cible vibratoire donnée en plaçant par exemple les pôles de (A + BC - BK), les fréquences et amortissements des résonances étant contrôlées par le vecteur K et les amplitudes par bande étant contrôlées par la matrice C.

The controlled box thus has the dynamics of (A + BC - BK) and more that of (A - BK) with the controller according to the first classic approach. The vector K is calculated for:

• provide stability for all changes to vector C,
• reach a given vibratory target by placing for example the poles of (A + BC - BK), the frequencies and damping of the resonances being controlled by the vector K and the amplitudes per band being controlled by the matrix C.

Of course, this is an exemplary embodiment to illustrate the characteristics taken into account at the CAP sensor, as illustrated in the figure 6 , directly for the CTL FF feedforward control, but indirectly also for the CTL FB feedback control and vice versa. Indeed, the feedforward control is considered here as applying a modification of the vibration characteristics to the sensor.

Referring now to the figure 7 summarizing an example of a succession of steps of a method within the meaning of the invention, at the end of a start step S1 aiming for example at the connection of the device DIS to the instrument/sensor/actuators system, it is measured, in practice, the transfer function H1 in open loop feedforward at step S2, which makes it possible to deduce at step S3 the parameters vibrations of the radiating structure and in particular the shape of the transfer function H1 and, from there, in step S4 the parameters of the feedback control. Then, at step S5, the musician can program a particular sound and/or effect setting, in which case the parameters of the feedforward control are updated at step S6, as well as the other parameters estimated at steps S3 and S4. In addition or as a variant, the sound adjustment can be carried out automatically, for example according to the particular attack of the musician, or other. Moreover, in one possible embodiment, the effect may not be chosen directly and restrictively by the musician, but may be programmed dynamically according to the playing of the musician.

Otherwise ("no" arrow at the output of the test S5), the device DIS can perform real-time processing in step S7 to apply the sound and/or effect setting programmed by the user, for playback at step S8 by the instrument itself.

Thus, the method above takes particular account of the feedforward control parameters in the estimation of the vibration parameters and the calculation of the feedback control gains.

The present invention then makes it possible to drastically reduce the instabilities and to obtain the sound level and more generally the targeted acoustic qualities, thanks to a hybrid feedback/feedforward controller, that is to say that the conventional digital audio effects and the processing of the intrinsic feedback to the instrument are calculated together to feed back the vibration signal to one or more ACT actuators of the radiating structure of the instrument.

• une augmentation du niveau sonore et un enrichissement du timbre de l'instrument acoustique,
• l'injection des traitements audionumériques dans un instrument acoustique en évitant les instabilités de type effet Larsen,
• l'atteinte des propriétés vibratoires de la structure rayonnante cibles que sont les fréquences, amortissements des résonances et amplitudes par bande fréquentielle, pour ainsi améliorer significativement les qualités acoustiques de l'instrument,
• un seul capteur et un seul actionneur pouvant être prévus pour effectuer la totalité des transformations.

Among the advantages of the technique implemented in the context of the present invention, mention may be made of:

• an increase in the sound level and an enrichment of the timbre of the acoustic instrument,
• the injection of digital audio processing into an acoustic instrument avoiding Larsen effect type instabilities,
• the achievement of the vibratory properties of the radiating structure targets which are the frequencies, damping of the resonances and amplitudes by frequency band, to thus significantly improve the acoustic qualities of the instrument,
• a single sensor and a single actuator can be provided to perform all the transformations.

Of course, the present invention is not limited to the embodiment described above by way of example; it extends to other variants.

Thus, a radiating structure has been described above, of the resonance box type of a stringed instrument (guitar type, or even violin or piano). However, the invention can also be applied to other musical instruments such as, for example, drum skins and drums, or even wind instruments. More generally still, the invention can be applied to any radiating structure (having a radiating plate or table coupled possibly but not necessarily to a sound box), or more generally to any electroacoustic system. It can be for example a loudspeaker, a computer case (or even a mobile device (smartphone or portable speaker) broadcasting sounds and music) conventionally having a sensor and a driven actuator within the meaning of the present invention.

Claims (9)

1. A method implemented by computer means, for processing sound data originating from at least one sensor mounted on an acoustically radiating structure of a music instrument and for activating at least one actuator applied against the acoustically radiating structure,

   the sensor (CAP) sensing an acoustic signal originating from the vibration of the radiating structure, said radiating structure carrying at least one actuator controlled by said computer means and involved in the vibration of the radiating structure,

   the method including:

   a) measuring a transfer function (H1) of the actuator, radiating structure, and sensor assembly,

   b) controlling activation of the actuator (ACT) to make the radiating structure vibrate, according to a selected setpoint:

   - while taking account of the measured transfer function, and

   - while taking account of the acoustic signal sensed by the sensor in a feedback mode, and wherein the activation of the actuator is controlled in a hybrid "feedback/feedforward" mode,

   and wherein the selected setpoint (S5) includes a command of at least one sound effect from among a sound amplitude change, an equalisation, an echo, a reverberation, a distortion, a phase change, a frequency change, an amplitude modulation, and a combination of these sound effects,

   the method including one or more iteration(s) of the steps:

   - on command of the sound effect selected by the user (S5), adjusting feedforward-type gains (S6) and gains of a feedback control (S4), according to the selected sound effect setpoint,

   - updating (S3) said transfer function measured with the adjusted gains,

   - controlling (S7) the activation of the actuator (ACT) to make (S8) the radiating structure vibrate, according to the setpoint corresponding to the selected sound effect, while taking account of the updated transfer function and of the acoustic signal sensed by the sensor in the feedback mode.
2. The method according to claim 1, wherein, in step a):

   - said transfer function (H1) is measured in open loop (S2), and

   - on this basis, vibratory parameters of the structure are estimated (S3) to calculate feedback control gains (S4).
3. The method according to one of the preceding claims, wherein the processing of the sound data is performed on a sample basis, at a latency lower than one hundred microseconds.
4. The method according to one of the preceding claims, wherein, the radiating structure including a sound box of a string musical instrument, said transfer function is measured with the strings being muted.
5. The method according to one of the preceding claims, wherein, the radiating structure including a sound box of a string musical instrument, two actuators are provided arranged on either side of the bridge carrying the strings.
6. The method according to claim 2, wherein a microphone is further provided to sense an acoustic pressure (p) in the air proximate to the radiating structure, the method further including measuring a second transfer function (H2) of the actuator, radiating structure, and microphone assembly,
   and wherein the activation of the actuator is controlled in a feedback/feedforward mode, with a refined estimation of the feedback control gains further based on said second transfer function (H2).
7. The method according to one of the preceding claims, wherein, the radiating structure including a sound box of a real musical instrument, said computer means are configured to confer vibratory and sound features of a selected musical instrument on said real musical instrument.
8. A computer program including instructions for the implementation of the method according to one of the preceding claims, when this program is executed by a processor (PROC).
9. A device (DIS) including a processing circuit configured for the implementation of the method according to one of claims 1 to 7.

Images