EP2898707B1 - Optimized calibration of a multi-loudspeaker sound restitution system - Google Patents

Description

The present invention relates to a method and a device for calibrating a sound reproduction system comprising a plurality of loudspeakers or sound reproduction elements. The calibration makes it possible to optimize the quality of listening of the restitution system that constitutes all the restitution elements, including the loudspeaker device and the listening room.

The restitution systems particularly concerned are the sound reproduction systems of the multichannel type (5.1, 7.1, 10.2, 22.2, etc.) or of the ambisonic type (Ambisonics in English or Higher Order Ambisonics (HOA)).

To allow good quality reproduction of multichannel signals, current devices for calibrating the acoustics of the listening environment are based on a general method of the "multichannel equalization" type in which the impulse responses of each loudspeaker of the restitution are measured using one or more microphones at one or more points of the listening location and a frequency equalization filtering is performed on each speaker, independently, by reversing all or part of the measured impulse response for the speaker concerned.

The inversion aims to correct the response of the loudspeaker so that it approaches as well as possible a "target" curve generally defined in the frequency domain to improve the rendering of the timbre of the sound sources.

Such a method is for example described in the document entitled " Digital Filter Design for Inversion Problems in Sound Reproduction ", by authors Kirkeby and Nelson, in JAES 7/8, pp.583-595, 1999 .

This type of calibration or correction focuses on the correction of the frequency aspect of the response of the system of reproduction of the listening location without exploiting the temporal information such as the phenomena of reflections and in particular the first reflections of the sound signals.

However, the first reflections of sound signals have a significant impact on the auditory perception of the restored sound signal.

In addition, the analysis of the impulse responses carried out in the existing calibration methods is of monophonic type, that is to say that it does not take into account either the spatial information of the reflections such as the direction of incidence. .

The absence of temporal and spatial data of the reflections, does not allow to take into account the role of these reflections on the perception of the direct wave of the sound signal by a listener, and thus to adjust the correction according to their effect specific. The quality of the sound signal reproduced and perceived by the listener is therefore not optimal.

The techniques of the state of the art are based on the application of correction filters on each of the channels of the multi-channel signal, that is to say that each speaker of the system of restitution is corrected individually without taking into account the whole network of speakers.

The document entitled "Validation of the correction of reflections on the basis of a representation in spherical harmonics" by Romain Deprez ET AL published at the 10th French Congress of Acoustics on April 12, 2010 (2010-04-12), pages 1-6 XP055062229 , describes a method of correcting signals feeding the speakers to minimize the effects of reflections generated by these speakers. This method does not take into account the auditory perception of sound reflections to adapt the sound reproduction system.

There is therefore a need to optimize the calibration carried out on multichannel audio signal restitution systems on the one hand to take into account the temporal and spatial properties of sound reflections which impact the auditory perception of direct waves, in order to adjust the 'processing effort according to the perceptibility of degradations and thus limit the audible artefacts likely to be generated by excessively constrained processing carried out in existing calibration methods; and on the other hand to use the different speakers jointly, in order to distribute the processing effort over all of the speakers.

The present invention improves the situation.

• obtention de réponses impulsionnelles multidirectionnelles des haut-parleurs de l'ensemble de restitution à la reproduction d'un signal audio de référence;
• analyse des réponses impulsionnelles multidirectionnelles obtenues, dans un domaine de représentation spatio-temporelle, sur au moins une fenêtre temporelle englobant les instants d'arrivée des premières réflexions du signal audio de référence reproduit pour déterminer un ensemble de caractéristiques des ondes directes et des premières réflexions comprenant au moins l'amplitude;
• comparaison de l'amplitude de chacune des réflexions à un seuil de perceptibilité déterminé en fonction de caractéristiques de l'onde directe et des premières réflexions du signal audio de référence et identification des réflexions non perceptibles pour lesquelles l'amplitude est inférieure au seuil déterminé;
• modification des réponses impulsionnelles obtenues pour obtenir des réponses impulsionnelles perceptives, par suppression des réflexions identifiées comme non perceptibles;
• détermination d'une matrice de filtrage par les étapes de :
  • détermination d'un signal d'erreur défini par la différence entre un signal de réponse cible d'un ensemble de restitution et un signal de réponse reconstruit à partir des réponses impulsionnelles perceptives;
  • inversion multicanale par minimisation du signal d'erreur ainsi déterminé pour obtenir les filtres de la matrice de filtrage,
  pour une application de cette matrice de filtrage au signal audio multi canal avant restitution sonore.

To this end, it proposes a method for calibrating a set of sound reproduction of a multi-channel sound signal comprising a plurality of loudspeakers. The process is such that it includes the following steps:

• obtaining multidirectional impulse responses from the loudspeakers of the reproduction unit to the reproduction of a reference audio signal;
• analysis of the multidirectional impulse responses obtained, in a space-time representation domain, over at least one time window including the arrival times of the first reflections of the reproduced reference audio signal to determine a set of characteristics of the direct waves and the first reflections comprising at least the amplitude;
• comparison of the amplitude of each of the reflections to a perceptibility threshold determined as a function of characteristics of the direct wave and of the first reflections of the reference audio signal and identification of the non-perceptible reflections for which the amplitude is less than the determined threshold;
• modification of the impulse responses obtained to obtain perceptual impulse responses, by suppression of the reflections identified as not perceptible;
• determination of a filtering matrix by the steps of:
  • determining an error signal defined by the difference between a target response signal from a rendering set and a response signal reconstructed from the perceptual impulse responses;
  • multichannel inversion by minimizing the error signal thus determined to obtain the filters of the filtering matrix,
  for an application of this filtering matrix to the multi-channel audio signal before sound reproduction.

Thus, in the implementation of the correction of the multi-channel audio reproduction system, the effect of the first reflections of the sound waves diffused by the reproduction system on the auditory perception of direct waves is evaluated and taken into account to adapt the processing. applied to the channels of the multi-channel signal according to the specific perceptual effect associated with each reflection. The filtering of the channels of the multi-channel signal thus takes into account exclusively the reflections which have an impact on the auditory perception of direct waves.

This therefore makes it possible to increase the quality of the audio signal reproduced.

In addition, as it is not necessary to take into account the reflections which are not perceptible, in the sense that their amplitude is less than a perceptible threshold, the constraints of the correction are lightened by the fact that they take into account perceptual impulse responses instead of raw impulse responses. In addition, some of the non-perceptible reflections which are eliminated from the impulse responses obtained correspond to components of the impulse response which are precisely at the origin of processing instabilities (in particular components with non-minimal phase). With perceptual impulse responses, the risks of instabilities and artefacts which can be generated during treatments taking into account all of the reflections are thus reduced.

The error signal thus determined makes it possible to take into account in the calculation of the filtering matrix, only the reflections which have an impact on the auditory perception of the direct wave. Indeed, only the reflections which are not perceptible are removed for the determination of the error signal.

The influence of reflections on the perception of the direct wave depends on several characteristics of the reflections. Advantageously, the perceptibility threshold can be obtained from characteristics determined by the step of analyzing the multidirectional impulse responses of the loudspeakers.

The various particular embodiments mentioned below can be added independently or in combination with one another, to the steps of the process defined above.

More particularly, the perceptibility threshold is determined as a function of the direction of incidence of the direct wave and / or of its amplitude, and of the directions of incidence of the first reflections and / or of their arrival times relative to the direct wave.

The effect of a reflection on the perception of the direct wave generally depends on five parameters in total; on the one hand it depends on two characteristics of the direct wave: its amplitude and its direction; on the other hand it depends on three characteristics of the reflection: its amplitude, its instant of arrival and its incidence.

However, if one of the characteristics of the direct wave is not known, it is possible to estimate the missing characteristic by fixing the other characteristic at an arbitrary value.

Similarly, if one of the pieces of information concerning the reflections is not known, one can for example estimate the perceptual effect of the reflection by fixing the missing characteristic at an arbitrary value, by taking for example the value corresponding to the case more unfavorable in order to increase the perceptibility. Thus, in the case where only the information of direction of the reflections is known, it is possible to fix a value to the characteristic of instant of arrival of the reflection in order to determine a value of the perceptibility threshold only with respect to the value direction, similarly if only the information of the time of arrival of the reflection is known, the direction value can be fixed and the perceptibility threshold determined only according to the value of the time of arrival. Finally, in the case where the two characteristics are known, the value of the threshold can be determined as a function of these two characteristics.

In one possible embodiment, the target response signal corresponds to the response of the direct wave alone without any reflection.

This allows to take into account as reference signal a signal devoid of any room effect.

In a first alternative embodiment, the target response signal corresponds to the response of a direct wave associated with reflections representative of a predetermined listening location.

The reference response can then be voluntarily chosen as a desired listening location in which the sound is at a desired quality.

In a second variant embodiment, the target response signal corresponds to the response of a direct wave associated with reflections representative of a different restitution set.

The reference response is here chosen as a function of a chosen reference rendering system, in which the number and the position of the loudspeakers can be different from the restitution system being corrected.

• un module d'obtention de réponses impulsionnelles multidirectionnelles des haut-parleurs de l'ensemble de restitution à la reproduction d'un signal audio de référence;
• un module d'analyse des réponses impulsionnelles multidirectionnelles obtenues, dans un domaine de représentation spatio-temporelle, sur au moins une fenêtre temporelle englobant les instants d'arrivée des premières réflexions du signal audio de référence reproduit pour déterminer un ensemble de caractéristiques des ondes directe et des premières réflexions associées comprenant au moins l'amplitude;
• un module de comparaison de l'amplitude de chacune des réflexions à un seuil de perceptibilité déterminé en fonction de caractéristiques de l'onde directe et des premières réflexions du signal audio de référence et d'identification des réflexions non perceptibles pour lesquelles l'amplitude est inférieure au seuil déterminé;
• un module de modification des réponses impulsionnelles obtenues pour obtenir des réponses impulsionnelles perceptives, par suppression des réflexions identifiées comme non perceptibles par le module d'identification;
• un module de calcul d'une matrice de filtrage apte à mettre en œuvre les étapes de:
  • détermination d'un signal d'erreur défini par la différence entre un signal de réponse cible d'un ensemble de restitution et un signal de réponse reconstruit à partir des réponses impulsionnelles perceptives;
  • inversion multicanale par minimisation du signal d'erreur ainsi déterminé pour obtenir les filtres de la matrice de filtrage ;
  pour une application de cette matrice de filtrage au signal audio multi canal avant restitution sonore.

The present invention also relates to a device for calibrating a set of sound reproduction of a multi-channel sound signal comprising a plurality of loudspeakers. This device is such that it includes:

• a module for obtaining multidirectional impulse responses from the loudspeakers of the reproduction unit for the reproduction of a reference audio signal;
• a module for analyzing the multidirectional impulse responses obtained, in a field of spatio-temporal representation, over at least one time window including the instants of arrival of the first reflections of the reproduced reference audio signal to determine a set of characteristics of direct waves and first associated reflections comprising at least the amplitude;
• a module for comparing the amplitude of each of the reflections at a perceptibility threshold determined as a function of characteristics of the direct wave and of the first reflections of the reference audio signal and of identifying non-perceptible reflections for which the amplitude is below the determined threshold;
• a module for modifying the impulse responses obtained in order to obtain perceptual impulse responses, by deleting the reflections identified as not perceptible by the identification module;
• a module for calculating a filtering matrix capable of implementing the steps of:
  • determining an error signal defined by the difference between a target response signal from a rendering set and a response signal reconstructed from the perceptual impulse responses;
  • multichannel inversion by minimizing the error signal thus determined to obtain the filters of the filtering matrix;
  for an application of this filtering matrix to the multi-channel audio signal before sound reproduction.

This device has the same advantages as the method described above, which it implements.

The invention also relates to an audio decoder comprising a calibration device as described.

It relates to a computer program comprising code instructions for implementing the steps of the calibration method as described, when these instructions are executed by a processor.

Finally, the invention relates to a storage medium, readable by a processor, integrated or not into the calibration device, possibly removable, storing a computer program implementing a calibration method as described above.

• la figure 1 représente un système de restitution sonore et un dispositif de calibration du système de restitution selon un mode de réalisation de l'invention;
• la figure 2 représente sous forme d'organigramme les étapes principales d'un procédé de calibration selon un mode de réalisation de l'invention;
• la figure 3a est une représentation d'un repère sphérique;
• la figure 3b, illustre les composantes harmoniques sphériques dans le cas d'une représentation spatiale ambisonique d'ordre 3;
• la figure 4 représente un exemple de tableau de valeurs en dB que peut prendre le seuil de perceptibilité utilisé dans le procédé de calibration selon un mode de réalisation de l'invention, pour un son direct d'angle d'incidence de 60°, en fonction de l'angle d'incidence (exprimé en degrés) de la réflexion et du temps d'arrivée (exprimé en ms) de cette réflexion par rapport à l'instant t0 d'arrivée de l'onde directe; le seuil de perceptibilité est défini comme le niveau (en dB) de la réflexion auquel est soustrait le niveau (en dB) de l'onde directe ;
• la figure 5 propose une autre illustration des valeurs prises par le seuil de perceptibilité : le seuil est cette fois représenté en fonction de l'incidence de la réflexion, et ceci pour différentes directions de l'onde directe ; dans tous les cas, le retard de la réflexion par rapport à l'onde directe est fixe et vaut 15 ms;
• la figure 6 représente un exemple d'une réponse impulsionnelle d'un haut-parleur d'un système de restitution ; le seuil de perceptibilité associé à chaque réflexion est également reproduit par une courbe pointillée;
• la figure 7 représente un exemple de réalisation matérielle d'un dispositif de calibration selon un mode de réalisation de l'invention.

Other characteristics and advantages of the invention will appear more clearly on reading the following description, given solely by way of nonlimiting example, and made with reference to the appended drawings, in which:

• the figure 1 represents a sound reproduction system and a device for calibrating the reproduction system according to an embodiment of the invention;
• the figure 2 represents in the form of a flowchart the main steps of a calibration method according to an embodiment of the invention;
• the figure 3a is a representation of a spherical coordinate system;
• the figure 3b , illustrates the spherical harmonic components in the case of an ambisonic spatial representation of order 3;
• the figure 4 represents an example of a table of values in dB that the perceptibility threshold used in the calibration process can take according to an embodiment of the invention, for a direct sound with an angle of incidence of 60 °, as a function of the angle of incidence (expressed in degrees) of the reflection and the arrival time (expressed in ms) of this reflection relative to the time t0 of arrival the direct wave; the perceptibility threshold is defined as the level (in dB) of the reflection from which the level (in dB) of the direct wave is subtracted;
• the figure 5 offers another illustration of the values taken by the perceptibility threshold: the threshold is this time represented as a function of the incidence of the reflection, and this for different directions of the direct wave; in all cases, the delay of the reflection with respect to the direct wave is fixed and is worth 15 ms;
• the figure 6 represents an example of an impulse response from a speaker of a reproduction system; the perceptibility threshold associated with each reflection is also reproduced by a dotted curve;
• the figure 7 shows an example of a hardware embodiment of a calibration device according to an embodiment of the invention.

The figure 1 therefore illustrates an example of a sound reproduction system in which the calibration method according to an embodiment of the invention is implemented. This system includes a processing device 100 comprising a calibration device E according to an embodiment of the invention controlling a rendering unit 180 which comprises a plurality of rendering elements (speakers, loudspeakers, etc.) represented here by loudspeakers HP ₁ , HP ₂ , HP ₃ , HP _i and HP _N.

These loudspeakers are arranged in a listening location in which a microphone or set of MA microphones is also provided.

These speakers and microphones are controlled by a processing device 100 which can be a decoder such as a living room decoder of the "set top box" type for reading or broadcasting audio or video content, a processing server capable of processing audio and video content and to retransmit them to the restitution unit, a conference bridge capable of processing the audio signals from different conference locations or any multi-channel audio signal processing device.

The processing device 100 comprises a calibration device E according to an embodiment of the invention and a filtering matrix 170 composed of a plurality of processing filters which are determined by the calibration device according to a calibration method such that '' illustrated later with reference to figure 2 .

This filtering matrix receives as input a multi-channel signal Si and transmits as output the signals SC ₁ , SC ₂ , SC _i , SC _N capable of being restored by the restitution assembly 180.

The calibration device E comprises a reception and transmission module 110 capable of transmitting on the one hand reference audio signals (Sref) to the various speakers of the reproduction unit 180 and to receive by the microphone or the 'set of microphones MA, the multidirectional impulse responses (RIs) of these different speakers corresponding to the broadcasting of these reference signals.

A multidirectional impulse response contains temporal and spatial information relating to all of the sound waves induced by the loudspeaker considered in the reproduction room.

The reference signals are for example signals whose frequency increases logarithmically over time, these signals being called in English "chirps" or "logarithmic sweeps".

The convolution of the signal measured at the output of the loudspeaker with an inverse reference signal makes it possible to directly obtain the impulse response of the loudspeaker.

In a particular embodiment adapted to the field of representation of spherical harmonics linked to the ambisonic or HOA format, the microphone capable of measuring the multidirectional impulse responses of the loudspeakers is a HOA type microphone placed at a point of the listening place, for example in the center of the loudspeakers of the reproduction unit.

This microphone will receive, for each speaker reproducing a reference audio signal, the sound reproduced in several directions. Indeed, the HOA microphone is made up of a plurality of microphones. By appropriate processing, the spatial information of the different sounds picked up can be extracted. For more details on this type of microphone, one can refer to the document entitled " Study and realization of advanced spatial encoding tools for the sound spatialization technique Higher Order Ambisonics: 3D microphone and distance control "by S. Moreau cited in Univ. Of Maine, PhD thesis, 2006 .

The HOA microphone then retrieves the multidirectional impulse responses from each of the speakers to transmit them to the calibration device or to store them in memory in a local or remote memory space.

When this information is stored in memory, the obtaining of these multidirectional impulse responses by the calibration device according to the invention is then carried out by a simple reading in memory.

These multidirectional impulse responses make it possible to obtain information on the directions of arrival of the direct waves and reflections of the restored signal as well as information on the time of arrival of both the direct waves and the reflections.

The analysis module 120 of the device E performs a joint analysis of the impulse responses obtained, which makes it possible to obtain these characteristics and in particular the characteristics of the first reflections of the restored signals. In the particular embodiment adapted to the field of representation of spherical harmonics, the multidirectional impulse responses are obtained in a spatio-temporal representation where the spatial information is described on the basis of the spherical harmonics and makes it possible to identify the directions of incidence different sound components. Thus, we obtain in the end all the information on the amplitude of the reflections, their directions of arrival and their arrival times in comparison to the arrival time of the direct wave. This step will be described later with reference to the figure 2 .

The analysis of impulse responses is done on a predetermined time scale, including the moments of the first reflections.

In an exemplary embodiment, this time window is between 50 and 100 ms in length, which corresponds to the time scale of the instants of arrival of the first reflections.

Of course, the embodiment thus described is adapted to the field of representation of spherical harmonics but it is entirely possible to carry out these same steps in a field of WFS representation (for "Wave Field Synthesis" in English) or in the plane wave domain. In these cases, the means for capturing the signals reproduced by the loudspeakers will have to be adapted to these areas of representation in order to obtain multidirectional impulse responses, without departing from the scope of the invention.

The calibration device E also includes a module 130 for comparing and identifying non-perceptible reflections. This module implements a step of comparing the amplitudes of the reflections, obtained by the analysis module 120, with a predetermined perceptibility threshold Se. This perceptibility threshold is determined by the module 140 from a predefined table of values stored in a memory space.

The determination of this perceptibility threshold will be explained later with reference to figures 4 and 5 .

In the case where the amplitude of a reflection is less than the perceptibility threshold as defined, this means that this reflection has no significant impact on the auditory perception of the direct wave of the restored signal.

A step of identifying these "non-perceptible" reflections is then implemented by the module 130. These identified reflections make it possible to implement by the module 150 a step of determining perceptual impulse responses which are deduced from the impulse responses obtained by module 110 by suppressing reflections judged as not perceptible.

Thus, only the reflections which have an impact on the perception of direct waves are taken into account to calculate in module 160, the filtering matrix Filt. of the matrix filtering module 170.

The figure 2 illustrates in the form of a flowchart, the main steps implemented in an embodiment of the calibration method according to the invention.

In step E201, the multidirectional impulse responses from the various loudspeakers of the reproduction unit as described with reference to the figure 1 , are obtained. They are obtained by the calibration device, either by simple reading in memory if these were saved beforehand, either by receiving the microphone or a set of microphones having carried out the measurement.

These multidirectional impulse responses are the responses of each loudspeaker following the reproduction of a reference signal as described with reference to the figure 1 .

A step E202 of analysis of the multidirectional impulse responses thus obtained is then implemented. This analysis is carried out in a space-time representation domain. Spatial information can for example be described in the field of representation of spherical harmonics. In this representation illustrated in figure 3a , each point has as spherical coordinates, a distance r from the origin 0, an angle θ of azimuth or orientation in the horizontal plane and an angle δ of elevation or orientation in the vertical plane. Preferably, the direction defined by (θ = 0 °, δ = 0 °) corresponds to the direction in front of the auditor. In such a frame, an acoustic wave is perfectly described if we define at any point at each time t, the acoustic pressure noted p (r, θ, δ, t) whose temporal Fourier transform is noted P (r, θ, δ, f) where f denotes the time frequency.

qui correspondent à la décomposition de l'onde de pression acoustique p sur la base des harmoniques sphériques. Par exemple, pour une source sonore en champ lointain, c'est-à-dire une onde plane d'incidence (θ_S, δ_S) portant un signal S(t), les composantes ambisoniques $B_{\mathit{mn}}^{\sigma }$

sont données par: $B_{\mathit{mn}}^{\sigma }=S\left(t\right).Y_{\mathit{mn}}^{\sigma }\left({\theta }_s,{\delta }_s\right)$

où les fonctions harmoniques sphériques $Y_{\mathit{mn}}^{\sigma }\left(\theta ,\delta \right)$

décrivent une base orthonormée: $$\begin{array}{r}{Y}_{\mathit{mn}}^{\sigma }\left(\theta ,\delta \right)=\sqrt{\left(2m+1\right)\left(2-{\delta }_{0,n}\right)\frac{\left(m-n\right)!}{\left(m+n\right)!}}{P}_{\mathit{mn}}\left(\sin \delta \right)\\ \times \{\begin{array}{ll}\cos \mathit{n\theta }& \mathrm{si}\sigma =+1\\ \sin \mathit{n\theta }& \mathrm{si}\sigma =-1\left(\mathrm{ignoré}\mathrm{si}n=0\right)\end{array}\end{array}$$

Les P_mn (sin δ) sont les fonctions de Legendre associées.In the context of higher order ambisonic spatialization (HOA), the spatial components are ambisonic components $B_{\mathit{min}}^{\sigma }$

which correspond to the decomposition of the sound pressure wave p on the basis of spherical harmonics. For example, for a far field sound source, i.e. a plane wave of incidence (θ _S , δ _S ) carrying a signal S (t), the ambisonic components $B_{\mathit{min}}^{\sigma }$

are given by: $B_{\mathit{min}}^{\sigma }=S\left(t\right).Y_{\mathit{min}}^{\sigma }\left({\theta }_s,{\delta }_s\right)$

where the spherical harmonic functions $Y_{\mathit{min}}^{\sigma }\left(\theta ,\delta \right)$

describe an orthonormal basis: $$\begin{array}{r}{Y}_{\mathit{min}}^{\sigma }\left(\theta ,\delta \right)=\sqrt{\left(2m+1\right)\left(2-{\mathrm{<figure-callout\; id="0"\; label="\delta "\; filenames="imgf0004.png,imgf0005.png"\; state="\{\{state\}\}">\delta </figure-callout>}}_{0,\mathrm{not}}\right)\frac{\left(m-\mathrm{not}\right)!}{\left(m+\mathrm{not}\right)!}}{P}_{\mathit{min}}\left(\sin \delta \right)\\ \times \{\begin{array}{ll}\cos \mathit{n\theta }& \mathrm{if}\sigma =+1\\ \sin \mathit{n\theta }& \mathrm{if}\sigma =-1\left(\mathrm{ignored}\mathrm{if}\mathrm{not}=0\right)\end{array}\end{array}$$

The P _min (sin δ ) are the associated Legendre functions.

(désignée comme la « composante W » dans la terminologie ambisonique) correspondant à l'ordre 0, les composantes bidirectives $Y_{10}^1,$

$Y_{11}^1,$

$Y_{11}^{-1}$

(désignée respectivement comme les «composantes Z, X et Y » dans la terminologie ambisonique) correspondant à l'ordre 1, et les composantes des ordres supérieurs.An illustration of the spherical harmonic functions is represented in figure 3b . We can thus see the omnidirectional component $Y_{}^{}$${}_{00}^1$

(designated as the “W component” in ambisonic terminology) corresponding to the order 0, the bidirectional components $Y_{}^{}$${}_{10}^1,$

${\mathrm{<figure-callout\; id="11"\; label="Y"\; filenames="imgf0002.png"\; state="\{\{state\}\}">Y</figure-callout>}}_{11}^1,$

$Y_{11}^{-1}$

(designated respectively as the “Z, X and Y components” in ambisonic terminology) corresponding to order 1, and the components of higher orders.

A three-dimensional or "3D" spatial representation called "of order M ' comprises K = ( M +1) ² components whose triplets of indices { m, n, σ } are such that 0≤ m ≤ M , 0≤ n ≤ m , σ = ± 1. A two-dimensional or "2D" representation of order M includes a subset of these components, retaining only the indices m = n, ie K = 2 M +1 components.

définissent donc un spectre spatial.Decomposition on the basis of spherical harmonics can be considered as the dual transform between spatial coordinates and spatial frequencies. Components $B_{\mathit{min}}^{\sigma }$

therefore define a spatial spectrum.

For each loudspeaker, at the end of step E201, a multidirectional impulse response is obtained which is made up of K impulse responses corresponding to the K components of the chosen spatial representation. In the case of the representation of spherical harmonics, these are the K components on the K = 2M + 1 spherical harmonics considered. For the j th loudspeaker, the multidirectional impulse response which is associated with it thus consists of K elementary responses H _jl (t) where the index l locates the index of the spatial component and t corresponds to the time sample. Subsequently, h _j (t) denotes the vector of the K spatial components measured for the jth speaker: $${\mathrm{h}}_{\mathrm{j}}\left(\mathrm{t}\right)=\left[\begin{array}{lllll}{\mathrm{H}}_{\mathrm{j}1}\left(\mathrm{t}\right)& \dots & {\mathrm{H}}_{\mathrm{jl}}\left(\mathrm{t}\right)& \dots & {\mathrm{H}}_{\mathrm{jK}}\left(\mathrm{t}\right)\end{array}\right].$$

If the reproduction system comprises a total of N loudspeakers, the set of multidirectional impulse responses measured for the N loudspeakers and the K spatial components defines a matrix H of size KxN, in which the jth column corresponds to the impulse response multidirectional associated with the jth speaker.

For each loudspeaker, the K spatial components contained in the vector h _j (t) represent the spatial spectrum of the sounds picked up by the microphone. To access the sound direction information, it is therefore necessary to perform an inverse transformation to go from a representation as a function of spatial frequencies to a representation as a function of spatial coordinates.

This inverse transformation is carried out by reconstructing the pressure wave p (r, θ, δ, t) by linear combination of spherical harmonics, each harmonic being weighted by the amplitude of the component associated with it. These elements are found in the thesis of S. Moreau cited above.

We can then evaluate the pressure wave p (r, θ, δ, t) at any point of a sphere centered on the point of measurement of multidirectional impulse responses by reconstructing the pressure wave point by point by linear combination of spherical harmonics. We can for example evaluate this pressure on a network of P points defining a "regular sampling" of the sphere in the sense defined in the dissertation of S. Moreau. This operation is similar to the spatial decoding of the ambisonic components for restitution by a regular spherical network of P virtual loudspeakers. This spatial decoding step is for example described in the document entitled " Ambisonics encoding of other audio formats for multiple listening conditions "by the authors Jérôme Daniel, Jean-Bernard Rault and Jean-Dominique Polack in AES 105th Convention, September 1998 .

pour les P directions des haut-parleurs virtuels, en regroupant les azimuths θ_q et élévations δ_q dans un unique doublet C = (θ_q , δ_q ) associé à un haut-parleur (q désigne l'indice du haut-parleur). Dans la matrice Y, chaque colonne est constituée des valeurs des K harmoniques sphériques pour un haut-parleur donné. Au final, on obtient, pour chaque haut-parleur et chaque échantillon temporel t, un vecteur G_j(t) de longueur P décrivant la distribution spatiale des composantes sonores captées sur un réseau de P points définissant un échantillonnage régulier de la sphère: $$G_j\left(t\right)=Y^T{\mathrm{h}}_{\mathrm{j}}\left(\mathrm{t}\right)$$

In practice, this transformation of the spatial frequencies (ambisonic components) towards the spatial coordinates is carried out by multiplying, for each loudspeaker and each temporal sample t, the vector h _j (t) by a decoding matrix D. For example, the matrix D can be obtained as D = Y ^T , where the matrix Y is calculated by evaluating the K spherical harmonics $Y_{\mathit{min}}^{\sigma }\left(\theta ,\delta \right)$

for the P directions of the virtual speakers, by grouping the azimuths θ _q and elevations δ _q in a single doublet C = ( θ _q , δ _q ) associated with a speaker (q denotes the index of the speaker) . In the matrix Y, each column consists of the values of the K spherical harmonics for a given loudspeaker. Finally, we obtain, for each loudspeaker and each time sample t, a vector G _j (t) of length P describing the spatial distribution of the sound components picked up on a network of P points defining a regular sampling of the sphere: $$G_j\left(t\right)=Y^T{\mathrm{h}}_{\mathrm{j}}\left(\mathrm{t}\right)$$

du point pour lequel le maximum de G_j(t) est observé, et son amplitude correspond à l'amplitude de ce maximum A_Ri=G_j(t_i). Dans ce qui précède, l'indice i repère l'indice de la réflexion considérée. La précision d'estimation de ces caractéristiques dépend donc du nombre P de haut-parleurs virtuels utilisés pour cette analyse. Le premier échantillon temporel pour lequel on observe un maximum définit l'instant d'arrivée de l'onde directe. On a soin de relever aussi l'amplitude (A_D) et l'incidence de cette dernière (C_D = (θ_D, δ_D) où θ_D et δ_D définissent respectivement l'angle d'azimut et l'angle d'élévation repérant la direction de l'onde directe).The maximum of this function G _j (t) identifies a reflection. If G _j (t) has several maxima, these different maxima each identify a reflection. Thus, for each identified reflection, its characteristics are determined according to the following procedure: its arrival instant corresponds to the sample t _Ri = t for which it is identified, its incidence corresponds to the spatial coordinates $${\mathrm{VS}}_{\mathit{Ri}}=\left({\theta }_{\mathit{Ri}},{\delta }_{\mathit{Ri}}\right)=\left({\theta }_q,{\delta }_q\right)$$

of the point for which the maximum of G _j (t) is observed, and its amplitude corresponds to the amplitude of this maximum A _Ri = G _j (t _i ). In the foregoing, the index i identifies the index of the reflection considered. The precision of estimation of these characteristics therefore depends on the number P of virtual loudspeakers used for this analysis. The first time sample for which a maximum is observed defines the instant of arrival of the direct wave. Care is also taken of the amplitude (A _D ) and the incidence of the latter (C _D = (θ _D , δ _D ) where θ _D and δ _D define the azimuth angle and the angle d respectively elevation identifying the direction of the direct wave).

et le retard entre l'onde directe et la réflexion : $${\tau }_{\mathit{Ri}}\left(j\right)=T_{\mathit{Ri}}\left(j\right)-T_D\left(j\right).$$

Thus, from the multidirectional impulse responses obtained, considered on a time analysis window including the instants of the first reflections of the audio signal reproduced by the loudspeakers, it is possible to determine, and this for each loudspeaker, the characteristics of the direct wave and the characteristics of the reflections associated with it. Thus, for the jth loudspeaker, the characteristics of the direct wave are determined on the one hand, such as its amplitude A _D (j), its instant of arrival on the microphone T _D (j) or its direction of incidence. C _D (j); and on the other hand the characteristics of the reflections such as their amplitudes A _Ri (j), their instants of arrival on the microphone T _Ri (j) or their directions of incidences C _Ri (j). In the following, we will rather use the amplitude normalized by the amplitude of the direct wave: $${\mathit{YEAR}}_{\mathit{Ri}}\left(j\right)=\frac{{\mathrm{AT}}_{\mathit{Ri}}\left(j\right)}{{\mathrm{AT}}_D\left(j\right)},$$

and the delay between the direct wave and the reflection: $${\tau }_{\mathit{Ri}}\left(j\right)=T_{\mathit{Ri}}\left(j\right)-T_D\left(j\right).$$

The first reflections of a reproduced audio signal depend on the listening location in which the reproduction unit is placed. Generally, these first reflections appear in a time located in a range from 50 to 100 ms after the direct wave.

Advantageously, the analysis time window of step E202 will, in an adapted embodiment, be between 50 and 100 ms.

Step E203 compares the amplitudes obtained by the analysis step with a perceptibility threshold Se of the reflections which has been defined beforehand and stored in memory. Step E204 makes it possible to find the predefined threshold value as a function of characteristics of each reflection and of the associated direct wave, obtained in the analysis step E202.

Indeed, several scenarios can arise. In a first embodiment, only the information on the direction of the reflections is known and recovered from the analysis step. To find the corresponding perceptibility threshold, the value of the characteristic of the time of arrival of the reflection is fixed, for example the most critical value (the one which gives maximum perceptibility) and the value of the perceptibility threshold is determined only relative to the value of leadership.

Likewise if only the information of time of arrival of the reflection is known, the direction value can be fixed, for example the most critical value (the one which gives maximum perceptibility), and determine the perceptibility threshold according to the value of the arrival time.

Finally, in the case where the two characteristics are known, the threshold value can be determined, with better precision, as a function of these two characteristics.

For this, an array of perceptibility threshold values is stored in memory. An example of such a table is illustrated with reference to the figure 4 . This table shows, for a direct sound located at an azimuth angle of 60 °, the value of the perceptibility threshold of a reflection expressed in dB, as a function of the angle of incidence characteristics of the reflection (ie its angle azimuth θ _Ri in the horizontal plane corresponding to the elevation δ _Ri = 0 °) and arrival time of this reflection with respect to the arrival time of the direct wave τ _Ri ( j ) . The threshold is defined as the relative level of the reflection, that is to say it represents the difference between the amplitude values (expressed in dB) of the reflection and of the direct wave considered.

This table of values is an example of threshold values defined from psycho-acoustic experiences carried out by considering different types of sound signal (speech, clicks, music, etc.), different angles of incidence and different times of arrival of reflections and of the direct wave. A threshold of perceptibility of these reflections is defined according to these parameters.

To complete the illustration of the perceptibility threshold values of the figure 4 , the figure 5 shows different perceptibility threshold curves expressed in dB (which always corresponds to the relative threshold corresponding to the difference between the level of the reflection and that of the direct wave). These different curves correspond to different positions of the direct wave (azimuth of 0 ° for D1, 60 ° for D2, 90 ° for D3 and 150 ° for D4) and represent the thresholds of perceptibility as a function of the direction of reflection, this for a fixed arrival time (corresponding in this case to 15 ms).

Thus, in step E204, the threshold value corresponding to the characteristics obtained in the analysis step is recovered. This threshold value is compared to the amplitude value of each reflection in step E203. To be compared to the perceptibility threshold, the value of the amplitude of the reflection is referenced to that of the associated direct wave and expressed in dB: 20log ( AN _Ri ( j )).

In the case where the amplitude value of the reflection is lower than the perceptibility threshold value, this means that this reflection has no impact on the perception that a listener of the direct wave can have. This reflection is therefore not to be taken into account for the processing of a multi-channel signal before restitution. Step E203 thus makes it possible to identify all the reflections which have no impact on the perception of the direct wave. Step E203 therefore identifies all the reflections for which the amplitude is below the perceptibility threshold.

To illustrate this step E203, the figure 6 represents an example of impulse response, for a given direction, of one of the loudspeakers of the reproduction unit in comparison with the curve in broken line representing the perceptibility threshold (RMT for "Reflection Masked Threshold") obtained by the table described above with reference to the figure 4 . Reflections whose level is lower than the threshold curve are thus identified. Note that in the illustrated case, the first reflections occurring in the first 15 ms are not perceptible.

From this identification of non-perceptible reflections, step E205 modifies the impulse responses h _j (t) obtained in step E201 for j = 1 to N loudspeakers, to obtain perceptual impulse responses hp _j (t). For this, the modification consists in eliminating the non-perceptible reflections identified in step E203 in the impulse responses.

In more detail, this operation is carried out for example by a thresholding operation. At each instant t, the value of the perceptibility threshold Se is subtracted from the impulse response signal which was obtained in step E201.

Preferably, this treatment is applied to the spatial spectrum defined by the K components h _j (t) = [H _j1 (t) ... H _jl (t) ... H _jK (t)] in the area of spatial representation. chosen, corresponding for example to the representation on the basis of spherical harmonics. However, the processing can also be applied in the dual domain of space coordinates. In the following, we will describe the operation performed in the case of the spatial spectrum.

The thresholding operation consists in comparing for each identified reflection its amplitude with the perceptibility threshold Se associated with its characteristics. Thus, for the ith reflection identified for the jth speaker, the threshold Se (i) is determined as a function of its characteristics [ τ _Ri ( j ), C _Ri (j)]. This reflection is localized at time t _i given by: $${\mathrm{t}}_i={\mathrm{T}}_{\mathrm{D}}\left(\mathrm{j}\right)+{\tau }_{\mathit{Ri}}\left(j\right).$$

$${\mathrm{HP}}_{\mathrm{jl}}\left({\mathrm{t}}_{\mathrm{i}}\right)=\left(\left|{\mathrm{H}}_{\mathrm{jl}}\left({\mathrm{t}}_{\mathrm{i}}\right)-{10}^{0.05\mathrm{Se}}\right|\right)\frac{{\mathrm{H}}_{\mathrm{jl}}\left({\mathrm{t}}_{\mathrm{i}}\right)}{\left|{\mathrm{H}}_{\mathrm{jl}}\left({\mathrm{t}}_{\mathrm{i}}\right)\right|}\mathrm{si}{\mathrm{AN}}_{\mathrm{Ri}}\left(\mathrm{j}\right)>{10}^{0.05\mathrm{Se}}$$

où HP_jl(t) désigne la réponse impulsionnelle perceptive associée à H_jl(t).To carry out the thresholding, we therefore consider the impulse response at this instant, either h _j (t _i ), or more exactly on the associated spatial spectrum consisting of the K components [H _j1 (t _i ) ... H _jl (t _i ) ... H _jK (t _i )]. Several strategies are then possible. The simplest consists in preserving the relative amplitude of the components of the spatial spectrum, that is to say that identical processing is applied to all the components. In this case, for each component H _jl (t _i ), the thresholding operation can result in the following equations: $$\begin{array}{cc}{\mathrm{HP}}_{\mathrm{jl}}\left({\mathrm{t}}_{\mathrm{i}}\right)=0& \mathrm{if}{\mathrm{YEAR}}_{\mathrm{Ri}}\left(\mathrm{j}\right)\le {10}^{0.05\mathrm{Se}}\end{array}$$

$${\mathrm{HP}}_{\mathrm{jl}}\left({\mathrm{t}}_{\mathrm{i}}\right)=\left(\left|{\mathrm{H}}_{\mathrm{jl}}\left({\mathrm{t}}_{\mathrm{i}}\right)-{10}^{0.05\mathrm{Se}}\right|\right)\frac{{\mathrm{H}}_{\mathrm{jl}}\left({\mathrm{t}}_{\mathrm{i}}\right)}{\left|{\mathrm{H}}_{\mathrm{jl}}\left({\mathrm{t}}_{\mathrm{i}}\right)\right|}\mathrm{if}{\mathrm{YEAR}}_{\mathrm{Ri}}\left(\mathrm{j}\right)>{10}^{0.05\mathrm{Se}}$$

where HP _jl (t) denotes the perceptual impulse response associated with H _jl (t).

Thus, the perceptual impulse responses only retain the reflections having a significant impact on the perception of the direct wave.

These perceptual impulse responses are then used to determine the filtering matrix, in step E206. This filtering matrix is then used to process the multi-channel audio signal before its sound reproduction by the system reproduction unit.

To obtain the set of filters constituting the Filt filtering matrix of the processing device, a possible embodiment includes a step of determining an error signal defined by the difference between a predetermined target response signal from the set of restitution and a response signal reconstructed from the perceptual impulse responses and a multichannel inversion step by minimization of the error signal thus determined.

The error signal thus obtained therefore only takes into account the perceptible reflections since it is calculated from a reconstructed signal based on the perceptual impulse responses.

The inversion can be performed by a gradient descent algorithm or its variants. An example of a possible inversion algorithm is that of the ISTA type (for "Iterative Shrinkage-Thresholding algorithm) as described in the document entitled" A Fast Iterative Shrinkage-Thresholding Algorithm for Linear Inverse Problems "by the authors Amir Beck & Marc Teboulle, published in SIAM J. IMAGING SCIENCES, Vol. 2, No. 1, pp. 183-202 in 2009 .

In general, the problem which arises to calculate the filters of the processing matrix is as follows. There are N loudspeakers which constitute the real reproduction system. In the context of higher order ambisonic spatialization (HOA), the space of spatial representation is of dimension K. Spatial information is therefore described by K coefficients. The objective is to reproduce with the N speaker system, a set of V signals defining the input multichannel audio signal. These V signals are dedicated to an ideal reproduction system consisting of V loudspeakers. This ideal system defines the V target signals which one wishes to reproduce and which therefore correspond to the responses of a fictitious system of V virtual loudspeakers. In the simplest case, the actual reproduction system also includes N = V loudspeakers. But in the general case, we are able to emulate a system of V virtual speakers from a device of N real speakers.

• avec H, la matrice de dimension KxN comportant les réponses impulsionnelles des N éléments du système de restitution dans le domaine d'analyse spatiale,
• W, la matrice comportant les filtres de correction à calculer, de dimension NxV,
• T, la matrice contenant les V réponses cibles définies dans le domaine d'analyse spatiale, de dimension KxV,
• et l'opération dénotée par « * » est un produit matriciel convolutif où un élément T_ij de la matrice T est obtenu de la façon suivante : $$T_{\mathit{ij}}\left(t\right)={\displaystyle \sum _{k=1}^N{H}_{\mathit{ik}}\ast W_{\mathit{kj}}\left(t\right)}$$
  

Chaque matrice est une matrice de vecteurs, au sens où la troisième dimension correspond à l'échelle des temps.
L'objectif de l'opération d'inversion est de trouver les éléments de la matrice W.The equation to solve is as follows: $$\mathrm{T}\left(\mathrm{t}\right)=\mathrm{H}\ast \mathrm{W}\left(\mathrm{t}\right)$$

• with H, the matrix of dimension KxN comprising the impulse responses of the N elements of the restitution system in the field of spatial analysis,
• W, the matrix comprising the correction filters to be calculated, of dimension NxV,
• T, the matrix containing the V target responses defined in the area of spatial analysis, of dimension KxV,
• and the operation denoted by "*" is a convolutional matrix product where an element T _ij of the matrix T is obtained in the following way: $$T_{\mathit{ij}}\left(t\right)={\displaystyle \sum _{k=1}^{\mathrm{NOT}}{H}_{\mathit{ik}}\ast W_{\mathit{K\; J}}\left(t\right)}$$
  

Each matrix is a vector matrix, in the sense that the third dimension corresponds to the time scale.
The objective of the inversion operation is to find the elements of the matrix W.

The resolution of this operation can be carried out in two stages. First of all, the correction filters are calculated by correcting only the room effect of the restitution place, that is to say that we take into account the actual loudspeaker device, ie N loud- speakers. In a second step, the arrangement of the loudspeakers is compensated for in order to adapt the V signals to a restitution according to a non-ideal configuration of N loudspeakers. For this purpose, the V signals are distributed by matrixing on the N channels associated with the real reproduction system in order to emulate a system of V virtual loudspeakers.

In the present case, to implement the invention, the elements of the matrix H include the perceptual impulse responses as obtained in step E205.

The target responses may vary depending on the expected sound reproduction result.

In one embodiment, this target response corresponds to the impulse response given by the direct wave alone without any reflection. This is equivalent to suppressing all the room effect in the expected signal.

In a first alternative embodiment, the target response signal corresponds to the response of a direct wave associated with reflections representative of a predetermined listening location.

A characteristic listening location with good listening quality may be desired (for example the listening location in the Pleyel ™ room). In this case, the processing filters will be calculated to obtain a sound reproduction close to this listening quality.

In a second variant embodiment, the target response signal corresponds to the response of a direct wave associated with reflections representative of a restitution set different from that used to restore the resulting signal.

Thus, a desired restitution system, for example comprising more speakers, is taken as a reference for obtaining a restitution close to that which would have been obtained with such a system.

Other target response signals can obviously be chosen according to the effect of the desired restitution.

Thus, the implementation of the described method makes it possible to obtain a better quality of listening during the restitution of a multi-channel audio signal thanks to the taking into account only of the perceptible reflections of the signals by the restitution unit in the listening place.

The figure 7 shows an example of a hardware embodiment of a calibration device according to the invention. This can be an integral part of an audio / video decoder, a processing server, a conference bridge or any other audio or video playback or broadcasting equipment.

This type of device comprises a µP processor cooperating with a memory block MEM comprising a storage and / or working memory.

The memory block can advantageously include a computer program comprising code instructions for implementing the steps of the calibration method within the meaning of the invention, when these instructions are executed by the processor, and in particular the steps for obtaining responses. multidirectional impulse signals from the loudspeakers of the reproduction unit for reproducing a predetermined audio signal, for analyzing the multidirectional impulse responses obtained, in a space-time representation domain, over at least one time window including the instants d arrival of the first reflections of the predetermined audio signal reproduced to determine a set of characteristics of the first reflections, of comparing the amplitude of each of the reflections at a predetermined perceptibility threshold and of identifying the non-perceptible reflections for which the amplitude is less than the predetermined threshold, modification of the impulse responses obtained to obtain perceptual impulse responses, by deletion of the reflections identified as non-perceptible and determination of a filtering matrix from the perceptual impulse responses for an application of this filtering matrix to the multi-channel audio signal before sound reproduction.

Typically, the description of the figure 2 takes the steps of an algorithm of such a computer program. The computer program can also be stored on a memory medium readable by a reader of the device or downloadable in the memory space of the latter.

The memory MEM stores a table of perceptibility threshold values as a function of characteristics of the sound components consisting of the direct wave and of the reflections used in the method according to an embodiment of the invention and in general, all the necessary data. to the implementation of the process.

Such a device comprises an input module I capable of receiving impulse responses from a restitution assembly and an output module S capable of transmitting to a processing module, the filters calculated from a filtering matrix.

In one possible embodiment, the device thus described may also include the processing functions by implementing the processing matrix on reception at I of a multi-channel signal Si to output processed signals SCi capable of be returned by the return package.

Claims (9)

1. Method for calibrating an assembly for sound playback of a multi-channel sound signal having a plurality of loudspeakers, the method having the following steps:

   - obtaining (E201) multi-directional impulse responses from the loudspeakers of the playback assembly upon reproduction of a reference audio signal;

   - analyzing (E202) the multi-directional impulse responses obtained, in a domain of spatio-temporal representation, over at least one time window encompassing the instants of arrival of the early reflections of the reproduced reference audio signal in order to determine a set of characteristics of the direct waves and of the associated early reflections comprising at least the amplitude;

   the method being characterized in that is has the following steps:

   - comparing (E203) the amplitude of each of the reflections with a determined perceptibility threshold (E204) as a function of characteristics of the direct wave and of the early reflections of the reference audio signal and identifying (E203) the non-perceptible reflections for which the amplitude is below the determined threshold;

   - modifying (E205) the impulse responses obtained in order to obtain perceptual impulse responses, by suppression of the reflections identified as non-perceptible;

   - determining (E206) a filtering matrix with the steps of:

   - determination of an error signal defined by the difference between a target response signal for a playback assembly and a response signal reconstructed from the perceptual impulse responses;

   - multi-channel inversion by minimization of the error signal thus determined in order to obtain the filters of the filtering matrix,

   for an application of this filtering matrix to the multi-channel audio signal before sound playback.
2. Method according to Claim 1, characterized in that the perceptibility threshold is determined as a function of the direction of incidence of the direct wave (C_D) and/or its amplitude (A_D), and the directions of incidence of the early reflections (C_Ri) and/or their arrival times (τ_Ri) with respect to the direct wave.
3. Method according to Claim 1, characterized in that the target response signal corresponds to the response of the direct wave alone without any reflection.
4. Method according to Claim 1, characterized in that the target response signal corresponds to the response of a direct wave associated with reflections representing a predetermined listening site.
5. Method according to Claim 1, characterized in that the target response signal corresponds to the response of a direct wave associated with reflections representing a different playback assembly.
6. Device for calibrating an assembly for sound playback of a multi-channel sound signal having a plurality of loudspeakers, the device having:

   - a module (110) for obtaining multi-directional impulse responses from the loudspeakers of the playback assembly upon reproduction of a reference audio signal;

   - a module (120) for analyzing the multi-directional impulse responses obtained, in a domain of spatio-temporal representation, over at least one time window encompassing the instants of arrival of the early reflections of the reproduced reference audio signal in order to determine a set of characteristics of the direct waves and of the associated early reflections comprising at least the amplitude;

   the device being characterized in that it comprises:

   - a module (120) for comparing the amplitude of each of the reflections with a determined perceptibility threshold (140) as a function of characteristics of the direct wave and of the early reflections of the reference audio signal and for identifying (120) the non-perceptible reflections for which the amplitude is below the determined threshold;

   - a module (150) for modifying the impulse responses obtained in order to obtain perceptual impulse responses, by suppression of the reflections identified as non-perceptible by the identification module;

   - a module (130) for computing a filtering matrix able to implement the steps of:

   - determination of an error signal defined by the difference between a target response signal for a playback assembly and a response signal reconstructed from the perceptual impulse responses;

   - multi-channel inversion by minimization of the error signal thus determined in order to obtain the filters of the filtering matrix;

   for an application of this filtering matrix to the multi-channel audio signal before sound playback.
7. Audio decoder having a calibration device according to Claim 6.
8. Computer program having code instructions for the implementation of the steps of the calibration method according to one of Claims 1 to 5 when these instructions are executed by a processor.
9. Storage medium, readable by a processor, on which a computer program is stored comprising code instructions for the execution of the steps of the calibration method according to one of Claims 1 to 5.

Images