EP1886535B1 - Method of producing a plurality of time signals - Google Patents

Description

The invention essentially relates to a method for producing more than two distinct temporal electrical signals from a first and a second time electrical signal. The invention finds a particularly advantageous application in the field of sound processing, for transforming a stereophonic sound signal into a multichannel sound signal such as, for example, the so-called 5.1 system which is broadcast using at least five loudspeakers. . In a sound system broadcasting a 5.1 signal, each speaker is intended to broadcast a sound signal which is distinct from other broadcast signals.

In practice, the 5.1 signals are generally broadcast by audio systems arranged inside a cinema, an apartment or a car. Such systems provide a listener in the center of the five-speaker space with the feeling of being enveloped by rich sound from five different sources. Indeed, the simultaneous broadcast of the five or six separate signals by as many independent speakers gives the sound signal a certain wrap.

On the other hand, with a conventional stereophonic system, the listener does not have that feeling of enveloping and depth of sound. Indeed, the listener only has the impression that the sound is propagated from a plane passing through the speakers, because the number of signals and sound sources is generally limited to two in a stereophonic system.

One of the aims of certain existing methods is therefore to transform stereophonic sound signals into 5.1 sound signals, in order to obtain the best possible auditory comfort. A 5.1 signal is broadcast by a system with at least five speakers: a center speaker, two left and right speakers, and two left and right rear speakers. A sixth speaker can optionally be added to this device to manage the low frequencies.

In a first approach, to obtain a 5.1 signal from a stereophonic signal, it would be possible to duplicate the two stereophonic signals on the five loudspeakers. However, such duplication would not provide the desired feeling of wrap to the user. Indeed, even if one multiplies the number of sound sources, one does not multiply the number of distinct signals diffused which give this richness to the sound.

In other known methods, the monophonic components are separated from the stereophonic components contained within the sound signals of a stereophonic system and the corresponding signals are broadcast by means of five loudspeakers.

More precisely, in these methods, the monophonic components of the original stereophonic sound signals are detected and the corresponding signal is broadcast using the central loudspeaker. In addition, to produce front-end sound signals, the monophonic component is subtracted from the original sound signals and the sound signals obtained are broadcast using the front speakers. In order to produce the rear signals, the anti-phase components of the original sound signals are detected and the sound signals obtained are broadcast using the rear loudspeakers. Indeed, the signals in opposition of phase give the feeling that the diffused sound comes from the back or that it is further away from the point of listening than the other sounds. One of the challenges of the process is therefore to achieve a good sound discrimination between the different sound signals so that each speaker diffuse a particular sound of its own.

To produce these five sound signals, a method is known in which a filter is applied to the stereo and sound electrical sound signals. However, this temporal processing involves the use of compressors which have relatively long response times. These long response times cause a pumping, ie a sudden change in intensity especially on the left and right channels when the central monophonic signal passes from a high noise level to a low noise level. Indeed, the left and right frontal sound signals include the monophonic component which is strongly attenuated as soon as it becomes strong in the center and is highlighted as soon as it fades to the center. However, there is a certain inertia between the attenuation and the highlighting of the monophonic component. This inertia gives sensations of void sound at certain times.

In addition, this method does not provide a good stereo back. In fact, to obtain the rear signals, a same electrical sound signal is broadcast on the two rear loudspeakers. The rear signals thus comprise the components of the stereophonic signals in phase opposition, but are monophonic with each other.

A method is also known in which a sound signal is more clearly separated from another. For this purpose, one of the steps of this method consists in eliminating certain components of the signals obtained which are below a threshold. This step reduces a measured crosstalk between two adjacent speakers. This crosstalk characterizes the separation between two adjacent speakers. However, the pumping effect is still present.

Another method of sound transformation is known in which filtering is likely to adapt over time. However, this method has some instability. Indeed, over time, the sound sources around the listener seem to move. With such a method, it is not possible to obtain the same sound effect throughout the broadcast. This method thus gives an unpleasant sensation to the listener of sound variations and does not respect the sound effect desired by the creator of the original soundtrack.

Also known is a method in which a large reverberation is applied to stereo sound signals. This reverberation corresponds to increasingly dense echoes. The method thus provides some virtual enveloping effect but can not make the richness of the diffusion of five different sound signals around the listener. Indeed, all the sound signals comprise a common modification information. In this process, we do not really discriminate information between the five sound signals, but we play on the stamp of the pieces. As a result, here too, we change the nature of the original work broadcast.

The invention proposes, in particular, to achieve a better discrimination between the different sound signals, while solving these pumping problems and respect for the original work.

The explanations that follow are given for sound signals. The teaching of the invention is however applicable to other fields, particularly to the transport of any electrical signals.

For this purpose, in the invention, the transformations of stereophonic sound signals are essentially in the frequency domain. Indeed, in the invention, the temporal stereophonic electrical signals are converted into frequency stereophonic electrical signals. Next, the frequency components in phase and the out-of-phase frequency components are identified for broadcasting on the central loudspeaker and the rear loudspeakers, respectively.

More precisely, in the invention, in order to identify the in-phase components, a monophonic filter is created, the coefficients of which are developed in particular from the difference of the stereophonic frequency electrical signals, and this filter is applied to the sum of the frequency components of the signals. stereophonic

These in-phase frequency components are, moreover, subtracted from the frequency components of the electrical signals of sound stereophonic in order to obtain the left and right frontal sound signals.

To obtain the out-of-phase components, a stereophonic filter is created whose coefficients are elaborated, in particular, from the sum of the frequency components of the two stereophonic electrical signals, and this filter is applied to each of the frequency components of the stereophonic electrical signals.

The fact of using frequency signals makes it possible to obtain an excellent rejection of the monophonic component and thus to avoid the effects of pumping since it is no longer necessary to modulate the levels of the right and left signals in order to mask the component monophonic residual. In addition, the processing is very fast, and even if it had to be delayed or delayed, the delay applies to all the signals simultaneously. There is therefore no impression of variation of sound intensity. Furthermore, in the invention, only the discrimination of the different components of the stereophonic signals is concerned, without modifying the sound signals by introducing for example a virtual acoustic effect of the reverberation type. The work as such is therefore perfectly respected during its dissemination which is the one wanted by its creator.

In addition, the stereophonic reconstruction from the five electrical sound signals generated by the invention is perfect, that is to say that we find exactly the original signal, which is not the case of other known methods.

Alternatively, it is possible to apply a low-pass filter and a high-pass filter on the electrical signal of its central. It is thus possible to create a new bass sound source making the listener's sound space even richer.

In addition, the filter according to the invention for extracting the in-phase components can be used to transport N original signals via two transport signals. Indeed, by summing the N original signals with each transport signal, after modulating or delaying each one in a particular way, it is possible to find these N original signals by applying to the transport signals modulations or inverse delays of those applied. initially, and applying a monophonic filter on the transport signals thus put in phase.

• on produit, dans le domaine fréquentiel, un signal électrique fréquentiel de son central comportant des composantes fréquentielles à partir de composantes fréquentielles en phase, en particulier présentes dans des proportions voisines dans les signaux électriques temporels de son droit et gauche initiaux, et
• on convertit le signal électrique fréquentiel de son central en un signal électrique temporel de son central,
• on produit, un signal électrique temporel de son gauche frontal par soustraction du signal électrique temporel de son central du signal électrique temporel de son gauche initial,
• on produit un signal électrique temporel de son droit frontal par soustraction du signal électrique temporel de son central du signal électrique temporel de son droit initial.

The invention therefore relates to a method for producing more than two distinct temporal electrical sound signals from a temporal electrical signal of its initial right and of a temporal electrical signal from its initial left, characterized in that:

• in the frequency domain, an electrical frequency signal of its central sound is produced comprising frequency components from frequency components in phase, in particular present in neighboring proportions in the initial temporal electrical signals of its right and left, and
• the electrical frequency signal of its central is converted into a temporal electrical signal of its central,
• producing a temporal electrical signal from its front left by subtracting the temporal electrical signal from its central from the temporal electrical signal from its initial left,
• a temporal electrical signal is produced from its front right by subtracting the time electrical signal from its central from the temporal electrical signal of its initial right.

Of course, in fine, the electrical signals produced are broadcast acoustically. However, after this production and before this broadcast, they may undergo additional modifications.

In a field related to that of leisure-type broadcasting mentioned above, the invention makes it possible to contribute to a better intelligibility of messages in the field of hearing aids. In a particular example, two starting left and right temporal signals are used, the above transformation is applied, and all or some of the signals produced are recombined so that only two time signals can be heard and heard with the prosthesis earphones. . The starting electrical signals are either signals measured by microphones at the location of each of the prostheses, or two signals measured by two microphones at a single prosthesis. As a result, the name "left" and "right" essentially identifies the fact that the starting sounds are different (regardless of their place of origin). In this case, we manage to create with the invention a depth of sound in the ears of users. This depth increases the intelligibility of the messages transmitted.

• on module chacun de ces signaux par une première modulation de phase, par une première modulation d'amplitude, et on lui applique un premier retard, ces premières modulations et premier retard étant définis par des premiers paramètres, et on obtient un premier signal modulé,
• on module chacun de ces signaux par une deuxième modulation de phase, par une deuxième modulation d'amplitude, et on lui applique un deuxième retard, ces deuxièmes modulations et deuxième retard étant définis par des deuxièmes paramètres, et on obtient un deuxième signal modulé,
• on somme les premiers signaux modulés de chacun des N signaux électriques indépendants originels, et on somme les deuxièmes signaux modulés de chacun des N signaux électriques indépendants originels et on obtient respectivement le premier et le deuxième signal de transport.

In addition, the invention relates to a method for transmitting N independent and original electrical signals via two electrical transport signals, characterized in that, for each of the N original signals,

• each of these signals is modulated by a first phase modulation, by a first amplitude modulation, and a first delay is applied to it, these first modulations and first delay being defined by first parameters, and a first modulated signal is obtained,
• each of these signals is modulated by a second phase modulation, by a second amplitude modulation, and a second delay is applied to it, these second modulations and the second delay being defined by second parameters, and a second modulated signal is obtained,
• the first modulated signals of each of the N original independent electrical signals are summed, and the second modulated signals of each of the N original independent electrical signals are summed and the first and second transport signals are respectively obtained.

• Figure 1 : une représentation schématique d'un système à au moins cinq haut-parleurs mettant en oeuvre le procédé selon l'invention ;
• Figure 2a : une représentation schématique d'une cellule appliquée aux signaux de son stéréophoniques produisant le signal électrique de son central comportant les composantes en phase de ces signaux ;

• Figure 2b : des représentations de composantes fréquentielles de signaux observables à différents endroits de la cellule de la figure 2a ;
• Figure 3a : une représentation schématique d'une cellule appliquée aux signaux de son stéréophoniques produisant les signaux arrière comportant des composantes en opposition de phase de ces signaux ;
• Figure 3b : des représentations de composantes fréquentielles de signaux observables à différents endroits de la cellule de la figure 3a ;
• Figure 4a : des représentations graphiques d'un système à encodeur décodeur mettant en oeuvre le procédé selon l'invention pour faire transiter N signaux électriques sur deux signaux de transport ;
• Figure 4b : une représentation schématique d'un encodeur selon l'invention permettant de transformer N signaux électriques en deux signaux électriques de transport;
• Figure 4c : une représentation schématique du décodeur selon l'invention permettant de reconstruire les N signaux à partir des deux signaux de transport émis par l'encodeur.

The invention will be better understood on reading the description which follows and on examining the figures which accompany it. These figures are given for explanatory purposes but in no way limitative of the invention. These figures show:

• Figure 1 : a schematic representation of a system with at least five loudspeakers implementing the method according to the invention;
• Figure 2a : a schematic representation of an applied cell stereophonic sound signals producing the electrical signal of its central comprising the in-phase components of these signals;

• Figure 2b : representations of frequency components of observable signals at different locations in the cell of the figure 2a ;
• Figure 3a : a schematic representation of a cell applied to the stereophonic sound signals producing the back signals having counter-phase components of these signals;
• Figure 3b : representations of frequency components of observable signals at different locations in the cell of the figure 3a ;
• Figure 4a : graphic representations of a decoder encoder system implementing the method according to the invention for passing N electrical signals on two transport signals;
• Figure 4b : a schematic representation of an encoder according to the invention for transforming N electrical signals into two electrical transport signals;
• Figure 4c : a schematic representation of the decoder according to the invention for reconstructing the N signals from the two transport signals transmitted by the encoder.

The figure 1 shows a stereophonic apparatus 1 which emits a temporal electrical signal GI (t) from its initial left and an electrical time signal DI (t) of its initial right. This stereophonic system 1 may for example be a CD player or MP3 files of portable or fixed type, a television, a laptop, or a mobile phone. In the remainder of the document, a signal expressed in the time domain will be designated S (t) and a signal expressed in the frequency domain by S (v).

In a typical case, the initial electrical signals GI (t) and DI (t) would be applied respectively to inputs of the loudspeakers 2 and 3 to be broadcast. Now, here, these signals are applied across a system 4 to be transformed into at least five distinct electrical signals 5.1: an electric signal C (t) of its central, an electric signal GF (t) of its left frontal, an electric signal DF (t) of its front right, an electric signal GA (t) of its left rear and an electric signal DA (t) of its rear right respectively diffused by loudspeakers 5-9.

To obtain the signal C (t) of its central sound, the signal GI (t) electrical of its initial left and the signal DI (t) electrical of its initial right are applied across a cell 10, respectively via a connection 16 and a connection 17 connecting outputs of the apparatus 1 to inputs of the cell 10. This cell 10 produces, in the field frequency, the signal C (v) electrical frequency of its central, from phase frequency components of the signals GI (v) and DI (v) electrical of its right and left initial. This cell then transforms the signal C (v) into a signal C (t) observable on its output. This signal C (t) is applied to an input of the speaker 5 to be broadcast.

In order to produce the electrical signals GF (t) and DF (t) temporally of its left and front right sides, the signal GI (t) of its initial left and the signal DI (t) of its initial right are respectively applied to a terminal of a subtractor 11 and 12, via connections 18 and 19 connecting the outputs of the apparatus 1 and inputs of the subtractors 11 and 12. The signal C (t) of its central sound is applied to a terminal of this subtractor 11 and this subtracter 12, via two connections 20 and 21 connecting the output of the cell 10 to the subtractive inputs of the subtractors 11 and 12.

The cell 11 thus produces a temporal electrical signal GF (t) of its frontal left by subtraction of the temporal electrical signal C (t) from the central sound of the electric signal GI (t) from its initial left. And the cell 12 produces a time electric signal DF (t) of its front right by subtraction of the electric signal C (t) from its central signal DI (t) electrical of its initial right. These signals GF (t) and DF (t) are applied to inputs of the speakers 6 and 7 to be broadcast.

To produce the electric signal GA (t) of its left rear and the electric signal DA (t) of its rear right, the signals GI (t) and DI (t) electrical of its left and right initial are applied to the terminals of a cell 13, via connections 22 and 23 connecting outputs of the apparatus 1 to inputs of the cell 13. This cell 13 transforms the signals GI (t) and DI (t) into GI (v) signals and DI (v) frequency and produces, in the frequency domain, the electric frequency signal GA (v) of its left rear and the electric signal DA (v) of its rear right, respectively from the GI (v) and DI signals. (v). The signals GA (v) and DA (v) essentially comprise frequency components with out-of-phase frequency values. These out of phase frequency values are values for which the frequency components of the electric GI (v) signal of its initial left have a significant phase shift compared to those of the electrical signal DI (v) of its initial right.

The cell 13 then transforms the GA (v) and DA (v) signals obtained into time signals GA (t) and DA (t). These time signals GA (t) and DA (t) are applied to inputs of the loudspeakers 8 and 9 via connections 27 and 28 respectively connecting an output of the cell 13 to an input of the loudspeakers 8 and 9.

Alternatively, it is possible to further produce a bass signal B (t) by applying the central temporal electric signal C (t) at the input of a low-pass filter 14 via a connecting connection 24. the output of the cell 10 and the input of the filter 14. This signal B (t) can be applied to an input of a bass speaker 16 to be broadcast. The high frequency portion of the central electrical signal C (t) is filtered using a high pass filter. The observable signal at the output of this filter 15 is then applied to the input of the loudspeaker 5, via a connection connecting the output of the filter 15 to the input of the loudspeaker 5.

As a variant, in the method according to the invention, only certain signals C (t), GF (t), DF (t), GA (t) and DA (t) are produced, and they are combined with one another by subtraction. , addition or convolution before broadcasting. Indeed, it may be interesting to broadcast only some of these signals to create, for example, special sound effects.

The figure 2a shows a detailed schematic representation of cell 10 of the figure 1 to obtain the electric signal C (t) of its central from the signals GI (t) and DI (t) electrical left and right.

More precisely, these initial signals GI (t) and DI (t) are applied at the input of a Fourier transform cell 35 via the connections 16 and 17. This Fourier transform cell transforms the signals GI (t) and DI (t) time respectively in DI (v) and GI (v) frequency signals. On the figure 2b , are represented the first three frequency components v1, v2, v3 of the signals DI (v) and GI (v). The first, second and third components of the signal DI (v) respectively have an amplitude of 0.1; 0.6 and -0.3. The first, second and third components of the signal GI (v) respectively have an amplitude of 0.5; 0.6 and 0.6.

The signals DI (v) and GI (v) are applied at the input of a cell 36 via connections 41 and 42 connecting the outputs of the cell 35 to inputs of the cell 36. This cell 36 subtracts, component component, the frequency components of the electrical signal DI (v) of its initial right from those of the signal GI (v) electric sound of its initial left to obtain frequency components of difference. The cell 36 then calculates a frequency difference module for each difference component. Thus, at the output of the cell 36, the signal | GI (v) - DI (v) | is obtained.

The figure 2b show this signal | GI (v) - DI (v) |. It is thus observed that, for the phase components of the signals DI (v) and GI (v), such as the second components v2, the signal | GI (v) - DI (v) | is zero. And for the components of the signals DI (v) and GI (v) which are out of phase, such as the third components v3, the signal | GI (v) - DI (v) | has relatively large values. For the components v1 of GI (v) and DI (v), we obtain a component v1 of the signal | GI (v) - DI (v) | worth 0.4.

The signal | GI (v) - DI (v) | is applied at the input of a cell 37 via a connection 43 connecting the output of the cell 36 to the input of the cell 37. This cell 37 subtracts each difference frequency module from a threshold value K1 allowing to obtain frequency residuals of difference. As a variant, it is possible to define several thresholds K1-KN that are assigned to different frequency ranges. The creation of a threshold K1 allows, as we will see, to set a tolerance when extracting the signal C (v). The higher the threshold, the more we tolerate components that are not exclusively monophonic. The lower the threshold, the less we tolerate components that are not monophonic.

Cell 37 then normalizes the frequency residues by dividing them by the threshold value K1. We thus obtain, on the figure 2b , a value of 0.3 for the first normalized residue, a value of 1 for the second normalized residual and a negative value for the third normalized residual which is greater than the threshold value. The normalized residuals associated with the in-phase components of the signals DI (v) and GI (v) thus have the value 1 while the normalized residuals associated with the out-of-phase components of the signals DI (v) and GI (v) have a value less than 1 .

The values of these residues are then used as parameters to produce a monophonic filter 38 called HM (v). Indeed, the electrical signal corresponding to the standardized residues is applied to an input of the filter 38 via a connection 44 connecting the output of the cell 37 to the input of the cell 38.

To construct this HM filter (v), if a frequency module is greater than the threshold value K1, then the value 0 is assigned to the frequency component concerned. In the opposite case, the frequency component concerned is retained. Thus, the coefficient of the HM filter (v) corresponding to the third frequency components v3 of the GI (v) and DI (v) signals has a zero value. Whereas the coefficients of the filter corresponding to the frequency components v1 and v2 of the GI (v) and DI (v) signals are unchanged.

The monophonic HM (v) filter is then applied to a sum, component to component, of the frequency components of the electrical signal of its initial right DI (v) and those of the electrical signal of its initial left GI (v). For this purpose, the signals DI (v) and GI (v) are applied to inputs of an adder 39, via connections 45 and 46 connecting the outputs of the cell 35 to an input of the adder 39. The signal observable in FIG. output of the summator 39 is applied to the input of the cell 38, via a connection 47 connecting an output of the summator 39 to an input of the filter 38.

At the output of the filter 38, is therefore observable a signal HM (v) * (GI (v) + DI (v)) corresponding to the signal C (v) electrical frequency of its central. On the figure 2b the frequency signal C (v) thus has a third component v3 zero, a second component v2 equal to 1.2 and a first component v1 equal to 0.2. This signal C (v) mainly comprises the in-phase components of the GI (v) and DI (v) signals.

The signal C (v) is then applied at the input of a cell 40 of inverse Fourier transform, via a connection 48 connecting the output of the filter 38 to the input of the cell 40. This cell 40 produces thus the signal C (t) electrical time of its central. This signal C (t) can then be applied to an input of a loudspeaker to be broadcast.

It has been seen that in order to obtain temporal electrical signals GF (t) and DF (t) from its front right and left, the signal C (t) of its central signal is subtracted from the signals GI (t) and DI (t). Now, we notice that here, with an electric signal C (v) of its central sound having a first component of amplitude 0.2, we will obtain an electric signal DF (v) of its frontal right having a first negative component of value -0.1 and an electrical GF (v) signal of its frontal left, including a first value component 0.4.

However, in some implementations of the method according to the invention, it is undesirable to create a phase opposition between the front temporal signals GF (t) and DF (t) left and right which did not exist at the beginning. . To solve this undesired phase shift problem, the MIN minimum is taken between the frequency component of the electrical DI (v) signal of its initial right and the frequency component of the electric GI (v) signal of its initial left. This MIN minimum is then compared with the generated frequency component of the electrical signal C (v) of its central. If the generated frequency component of the electric signal C (v) of its central unit is greater than this minimum MIN, then this minimum is retained. In the opposite case, we keep the component.

Here, for the first component v1 of the signal C (v), the value 0.2 will therefore be replaced by MIN = 0.1. one then obtains a first component of the electrical signal DF (v) of its frontal right which is equal to 0 and a first component of the electric signal GF (v) of its left equal to 0.4. Similarly, the value of the second component of the signal C (v) is replaced by 0.6, in order to avoid a phase difference appearing between the electrical signals of its left and right front ends.

Alternatively, the frequency frequency residues are directly used as weighting coefficients in the HM filter (v).

The figure 3a shows a detailed schematic representation of cell 13 of the figure 1 which makes it possible to obtain the temporal electrical signals DA (t) and GA (t) from its rear from the initial electrical time signals GI (t) and DI (t).

More specifically, the electrical signals DI (t) and GI (t) of its left and right temporal are applied to two distinct inputs of a Fourier transform cell 51, via the connections 22 and 23. A GI signal (v) electrical frequency of its initial left and an electrical signal DI (v) frequency of its right are observable at the exit of this cell 51. figure 3b shows the signals DI (v) and GI (v). The signal DI (v) has three first components v1-v3 frequency respectively worth 0.5; 0.2 and 0.6. The signal GI (v) comprises three first components v1-v3 frequency respectively worth 0; -0.2 and 0.6.

The signals DI (v) and GI (v) are respectively applied to inputs of a cell 52, via two connections 53 and 54 connecting the outputs of the cell 51 to inputs of the cell 52. This cell 52 adds, component to component, the frequency components of the signal DI (v) its initial right to those of the electric signal GI (v) of its initial left to obtain frequency components sum. This cell 52 then calculates a frequency modulus of sum for each frequency component sum. This cell 52 thus makes it possible to identify the out-of-phase components in the initial electrical frequency signals GI (v) and DI (v). On the figure 3b it is thus seen that the signal | (GI (v) + DI (v) | corresponding to the sum modulus of the signals GI (v) and DI (v) gives a zero value for the out-of-phase components, such as the second components v2 GI (v) and DI (v) signals, and a high value for the in-phase frequency components of the GI (v) and DI (v) signals.

Then the signal | GI (v) + DI (v) | electrical output obtained from the cell 52 is applied to the input of the cell 55, via a connection 56 connecting the output of the cell 52 to the input of the cell 55. This cell 55 subtracts each frequency module from a threshold value K'1, so as to obtain frequency residuals sum. Here again, there may be several thresholds K'1-K'N, each threshold K'1-K'N corresponding to a particular frequency range. These thresholds K'1-K'N give to the extraction of the signals GA (v) and DA (v) a certain tolerance by allowing, as we will see, to preserve components which are not completely in opposition of phase with each other.

Then, the cell 55 normalizes the residues by dividing them by the threshold value K'1. Normalized components are thus obtained which are equal to 1 for the components of the signals DI (v) and GI (v) exactly in phase opposition, such as the second components v2, and negative normalized components for the in-phase components of the GI signals ( v) and DI (v), such as the third components v3.

The signal obtained at the output of the cell 55 is then applied at the input of two identical filters 59, 60 called HSG (v) and HSD (v), respectively via a first and a second connection 57, 58 connecting an output of the cell 55 to an input of the filters 59 and 60. Thus, from these normalized residuals, the coefficients of the stereophonic filters HS (v) will be able to be developed.

More precisely, to create each of these filters 59-60, the components of the normalized signal that are less than zero are suppressed. In other words: if a frequency module of the signal GI (v) and DI (v) is greater than the threshold value K1, then the value zero is assigned to the frequency component concerned. In the opposite case, the frequency component concerned is retained. The first and second coefficients of HS (v) are thus equal to the standardized residuals corresponding to them. The third coefficient of HS (v) corresponding to in-phase frequency components of the signals DI (v) and GI (v) is zero.

In a next step, the component-component stereo filters 59 and 60 are respectively applied to frequency components of the electrical DI (v) signal of its initial right and frequency components of the electric GI (v) signal of its initial left. Thus, the signals DI (v) and GI (v) are respectively applied at the input of the filters 59 and 60, via the connections 61 and 62 respectively connecting an output of the cell 51 to an input of the filters 59 and 60.

This results in signals DA (v) and GA (v) electrical frequencies of its right and left back mainly comprising frequency components out of phase with each other. These signals DA (v) and GA (v) respectively correspond to the signals HS (v) * DI (v) and HS (v) * GI (v).

In a subsequent step, the signals DA (v) and GA (v) are applied at the input of an inverse Fourier transform cell 63 via a connection 64 and 65 connecting an output of the filters 59 and 60 to an input of the cell 63. Electrical signals DA (t) and GA (t) of its right and left rear transposed in the time domain are thus observable at the output of the cell 63. These signals DA (t) and GA (t) ) can be applied as speaker input for broadcast.

In a subsidiary step, for each frequency component of the signals DA (v) and GA (v), it is tested whether its value is greater in absolute value than the minimum MIN 'in absolute value of the signal components DI (v) and GI (v). ) initial. In the case where this component value is greater than the minimum, the value of the component concerned is replaced by the minimum. In the opposite case, we keep the component.

On the figure 3b , the value 0.1 of the first component v1 of the signal DA (v) is greater than the minimum MIN 'of the value of the first component of the signals DI (v) and GI (v) which is zero. Therefore the value 0.1 of the first component of the electrical signal of its right rear is replaced by the value 0. The other values of the components v2 and v3 of the signals GA (v) and DA (v) are retained. By performing this step, it is thus possible to keep, in the electrical signals GA (v) and DA (v) of its rear only the components which are out of phase with each other.

As a variant, the sum frequency residues are used as weighting coefficients of the frequency components in each stereophonic HS (v) filter.

As a variant, the frequency components of the signal C (v) are subtracted from the frequency components of the signals GI (v) and DI (v) using subtracters 66 and 67. And the signals observable at the output of these subtracters 66 and 67 are applied to inputs of the cell 52 and to the inputs of the filters 59 and 60. Such a variant makes it possible to ensure that no frequency component in phase of the signals DI (v) and GI (v) will be present in the Rear DA (v) and GA (v) signals produced.

In a particular application of a two-speaker broadcast system, such as a computer, a television set or a mobile telephone, to give a listener a feeling of soundwidth, the electrical signals DF (t) and GF (t). From DF (t), we subtract a part of GF (t) and GF (t), we subtract a part of DF (t). To these new signals, the signal C (t) is then added. This gives two sum time signals and is broadcast using speakers.

The figure 4a shows a system 71 which implements a method of transmitting N original electrical signals S1 (t) -SN (t) and independent via two electric transport signals L (t) and R (t).

More precisely, the system 71 comprises an encoder 72 at the input terminals of which, the signals S1 (t) -SN (t) are applied. This encoder 72 applies different filters on these signals S1 (t) -SN (t) and combines them so that they are transformed into two transport signals L (t) and R (t).

These transport signals L (t) and R (t) are applied at the input of a decoder 75, via connections 73 and 74 interconnecting the outputs of the encoder 72 and the inputs of the decoder 75. decoder 75 applies inverse filters to those applied by the encoder 72 on the L (t) and R (t) signals. The decoder 75 then extracts the components frequency signals which are in phase, so that the N original signals S1 (t) -SN (t) are observable on its outputs.

The figure 4b shows a detailed schematic representation of the encoder 72 according to the invention. Only the first four signals are represented here. The processing performed on the N original signals is similar to that performed on the first two signals S1 (t) -S2 (t).

The encoder 72 modulates each of the signals S1 (t), S2 (t) by a first amplitude modulation G1, G2, and applies a first delay R1, R2 on each of these signals. This first modulation and this first delay are defined by first parameters: G1 and G2 can thus be multiplying coefficients or attenuators of a few decibels. While the delays R1, R2 may be worth a few milliseconds. A first modulated signal T [S1 (t)], T [S2 (t)] is then obtained which is applied to an input terminal of an adder 76.

The encoder 72 also modulates each of the signals S1 (t), S2 (t) by a second amplitude modulation G'1, G'2, and applies a second delay R'1, R'2 on each of these signals. S1 (t), S2 (t). This second modulation and second delay are defined by second parameters: G'1, G'2 can thus be multiplying coefficients or attenuators of a few decibels. While the delays R'1, R'2 may be worth a few milliseconds. A second modulated signal T '[S1 (t)], T' [S2 (t)] is then obtained which is applied to an input terminal of a second adder 77.

The first summator 76 is the sum of the first modulated signals T [S1 (t)], T [(S2 (t)] of each of the original independent electrical signals, and a first transport signal L (t) corresponding to this sum is thus observable at its exit.

The second summator 77 is the sum of the modulated second signals T '[S1 (t)], T' [(S2 (t)] of each of the original independent electrical signals A second transport signal R (t) corresponding to this sum is thus observable at its exit.

As a variant, the original signals S1 (t), S2 (t) are also modulated by a first phase modulation φ1 and a second phase modulation φ'1, respectively to obtain the first T [S1 (t)], T [ (S2 (t)] and second T '[S1 (t)], T' [(S2 (t)] signals.

Thus, the first and second signals are all delayed and modulated in phase and amplitude, the delay may be zero in some cases, as the phase shift. A signal applied as it is to an input of an adder thus has a zero phase shift and an amplitude modulation ratio equal to 1.

The figure 4c shows a detailed representation of a decoder according to the invention. The first and second transport signals L (t), R (t) are applied to inputs of the decoder 75, via the connections 73 and 74.

This decoder 75 demodulates the first transport signal L (t) by N (here N = 2) first amplitude demodulations 1 / G1, 1 / G2, and it applies first N delays. These first 2N demodulations and N first delays are defined by 2N first inverse parameters. Each of the first 3N inverse parameters correspond to the inverse or opposite parameters of the first and second parameters. Amplitude demodulation allows to recover the amplitude of the original signals while the introduced delays make it possible to recalibrate in time and put back in phase the original signals. For delays, either introduce the inverse delay of each original delay, or introduce the difference between the two original delays as is the case in the figure. Indeed, instead of introducing a delay -R1 in the signal L (t) and a delay -R'1 in the signal R (t), a single delay R'1-R1 is introduced into the signal L (t). ). It is the same for the delay R'2-R2. N first demodulated signals D1 (t) -D2 (t) are thus obtained.

In a similar manner, the decoder 75 demodulates the second transport signal R (t) by N second amplitude demodulations 1 / G'1, 1 / G'2, and applies N second delays. These N second demodulations and N second delays are again defined by 2N second inverse parameters. These second inverse parameters have inverse or opposite values to those of the first and second parameters, so as to recover the amplitude and the phase of the original signals. N second demodulated signals D 1 (t) -D 2 (t) are thus obtained.

Couples of these first 2N D1 (t) -D2 (t) and second D1 (t) -D'2 (t) demodulated signals are selected and combined in monophonic filters 78-79. In each of these monophonic filters 78-79, an original electrical signal S1 (t) -S2 (t) is reconstructed from components frequency in phase of the electrical transport signals.

For this purpose, the first D1 (t) and the second D 1 (t) demodulated signal are applied to the input terminals of the monophonic filter 78. After demodulation, the demodulated signals D1 (t) and D'1 (t) comprise frequency components which have the same amplitude, which are in phase and which correspond to the frequency components of the original signal S1 (t). By applying the filter 78 which extracts the frequency components in phase from the signals applied to it at input, the signal S1 (t) is found again. In the same way, to reconstruct the original signal S2 (t), the demodulated signals D2 (t) and D'2 (t) are applied at the input of the filter 79.

Alternatively, if phase modulations φ1, -φ'1 had been made on the original signals to carry them, we would introduce N first inverse phase demodulations on the first transport signal L (t) and N second reverse phase demodulations. on the second transport signal R (t). Thus, to reconstruct the original signal S1 (t), we could introduce a phase shift - φ1 on L (t) and a phase shift -φ'1 on R (t).

Claims (15)

1. A method for producing more than two different electric time sound signals (C(t), GF(t), DF(t) GA(t), DA(t)) from an initial left electric time sound signal (GI(t)) and an initial right electric time sound signal (DI(t)), characterised in that:

   - in the frequency field, a central electric frequency signal (C(v)) is produced, comprising frequency components (v1, v3) from in-phase frequency components of the initial left and right electric sound signals (GI(v), DI(v)), these in-phase components having amplitudes, the difference between which is less than a threshold (K1-KN),

   - the central electric frequency sound signal (C(v)) is converted into a central electric time sound signal (C(t)),

   - a front left electric time sound signal (GF(t)) is produced by subtracting the central electric time sound signal (C(t)) from the initial left electric sound signal (GI(t)), and

   - a front right electric time sound signal (DF(t)) is produced by subtracting the central electric time sound signal (C(t)) from the initial right electric sound signal (DI(t)).
2. A method according to claim 1, characterised in that:

   - in the frequency field, a rear left electric frequency sound signal (GA(v)) and a rear right electric sound signal (DA(v)) are produced from the initial left and right electric sound signals (GI(v), DI(v)) respectively,

   - these rear left and right sound signals (GA(v), DA(v)) essentially comprising out-of-phase frequency components (v1-v3),

   - these out-of-phase components being components for which the frequency components of the initial left electric sound signal (GI(v)) have a significant phase difference compared to those of the initial right electric sound signal (DI(v)).
3. A method according to claim 1 or 2, characterised in that to produce the central electric sound signal (C(v)):

   - a monophonic filter (HM(v)) is applied to a sum total, component by component, of the frequency components of the initial left electric sound signal (GI(v)) and of those of the initial right electric sound signal (DI(v)), and

   - in the monophonic filter (HM(v))

   - the frequency components of the initial right electric sound signal (DI(v)) are subtracted, component by component, from those of the initial left electric sound signal (GI(v)) to obtain the differential frequency components,

   - a differential frequency module is calculated for each differential frequency component,

   - each differential frequency module is subtracted from a threshold value (K1) to obtain the differential frequency remainders, and

   - the differential frequency remainders are used as weighting coefficients for the frequency components in the monophonic filter (HM(v)).
4. A method according to claim 3, characterised in that to produce the central electric sound signal (C(v)),

   - the remainders are standardised by dividing them by the threshold value (K1).
5. A method according to claim 3 or 4, characterised in that to produce the central electric sound signal (C(v)),

   - if a frequency module is greater than the threshold value (K1), the value of zero is applied to the frequency component in question.
6. A method according to one of claims 3 to 5, characterised in that, for a given frequency component of the central electric sound signal (C(v)),

   - the minimum (MIN) between the frequency component of the right electric sound signal (DI(v)) and the frequency component of the left electric sound signal (GI(v)) is taken, and

   - this minimum is compared to the frequency component produced from the central electric sound signal (C(v)), and

   - if the frequency component produced from the central electric sound signal (C(v)) is greater than this minimum (MIN), this minimum is retained, and

   - if the frequency component produced from the central electric sound signal (C(v)) is lower than this minimum (MIN), this component is retained.
7. A method according to one of claims 1 to 6, characterised in that to produce the rear left and right electric signals (GA(v), DA(v)),

   - monophonic filters (HS(v)) are applied, component by component, to the frequency components of the initial left electric sound signal (GI(v)) and to the frequency components of the initial right electric sound signal (DI(v)) respectively, and

   - in each monophonic filter (HS(v))

   - the frequency components of the initial left electric sound signal (GI(v)) are added, component by component, to those of the initial right electric sound signal (DI(v)) to obtain the sum frequency components,

   - a sum frequency module is calculated for each sum frequency component,

   - each sum frequency module is subtracted from a threshold value (K1) to obtain the sum frequency remainders, and

   - the sum frequency remainders are used as weighting coefficients for the frequency components in each monophonic filter (HS(v)).
8. A method according to claim 7, characterised in that to produce the rear left and right electric sound signals (GA(v), DA(v)),

   - the remainders are standardised by dividing them by the threshold value (K1).
9. A method according to claim 7 or 8, characterised in that to produce the rear left and right electric sound signals (GA(v), DA(v)),

   - if a frequency module is greater than the threshold value, the value of zero is applied to the frequency component in question.
10. A method according to one of claims 7 to 9, characterised in that

    - for each frequency component of the rear electric sound signals,

    - the value of this component is compared to the minimum of the frequency component values of the front left and right electric sound signals, and

    - if this value is greater than the minimum, the component in question is replaced by the minimum.
11. A method according to one of claims 7 to 10, characterised in that, before applying the monophonic filters (HS(v)),

    - the frequency components of the central electric sound signal (C(v)) are subtracted from the frequency components of the initial left and right electric sound signals (GI(v), DI(v)).
12. A method according to one of claims 1 to 11, characterised in that

    - a low frequency central electric sound signal (C(v)) is produced by applying a low frequency filter (14) to the frequency components of the central electric sound signal.
13. A method according to one of claims 1 to 12, characterised in that

    - some of the more than two time signals produced are combined in order to produce only two combined time signals.
14. A method for transmitting N original and independent electric signals (S1-SN) via the intermediary of two electric transport signals (L(t), R(t)), characterised in that, for each of the N original signals,

    - each of these signals (S1(t)-SN(t)) is modulated by a first phase modulation (ϕ1), by a first amplitude modulation (G1, G2), and a first delay (R1, R2) is applied, these first modulations and this first delay being defined by first parameters to obtain a first modulated signal (T[S1(t)], T[(S2(t)]),

    - each of these signals (S1(t)-SN(t)) is modulated by a second phase modulation (ϕ'1), by a second amplitude modulation (G'1, G'2), and a second delay is applied, these second modulations and this second delay being defined by second parameters to obtain a second modulated signal (T'[S1(t)], T'[(S2(t)]),

    - the first modulated signals (T[S1 (t)], T[(S2(t)]) of each of the N original independent electric signals are added together, and the second modulated signals (T'[S1(t)], T'[(S2(t)]) of each of the N original independent electric signals are added together to obtain the first and second transport signals (L(t), R(t)) respectively.
15. A method according to claim 14, characterised in that

    - the first and second transport signals (L(t), R(t)) are received,

    - the first transport signal (L(t)) is demodulated by N first phase demodulations (-ϕ1), by N first amplitude demodulations (1/G1, 1/G2), and N first delays are applied, these 2N first demodulations and N first delays being defined by 3N first opposite parameters to obtain N first demodulated signals, each of the 3N first opposite parameters being the opposite parameter to the first parameters,

    - the second transport signal is demodulated by N second phase demodulations (-ϕ'1), by N second amplitude demodulations (1/G'1, 1/G'2), and N second delays are applied, these 2N second demodulations and N second delays being defined by 3N second parameters to obtain N second demodulated signals,

    - couples of these 2N first and second demodulated signals are selected and combined in monophonic filters, and

    - in each of the monophonic filters

    - an original electric signal is reconstructed from in-phase frequency components of the electric transport signals.

Images