EP2571290B1 - Local sound field synthesis with a virtual scattering body - Google Patents

Description

The present invention relates to the reproduction or synthesis of a sound field in a limited audience area, in particular the sound field synthesis by scattering on a virtual scattering body and subsequent time reversal.

Numerous methods for audio reproduction or physical synthesis of a sound field are known.

Thus, stereophonic techniques of audio reproduction with the help of two or more speakers by level differences or differences in transit time creates a spatial sound impression in natural hearing. The desired spatial hearing impression arises only within a limited range of the so-called "sweet spot". In the past, a number of methods have been presented which deal with the tracking or enlargement of the sweet spot in stereophonic (illusory) playback. Exemplary here on the pamphlets DE-A-102005052904 . DE-A-10125229 or US-B-6633648 directed.

In general, stereophonic sound can only convey the impression of a sound source that is at least as far away from the nearest speakers as possible. However, no physical synthesis of a desired complex sound field can be achieved.

Wave field synthesis (WFS), on the other hand, allows the physical synthesis of a sound field over a broad range. For example, a description of the wave field syn thesis can be found in the documents AJ Berkhout, D. de Vries, and P. Vogel. Acoustic control by wave field synthesis. Journal of the Acoustical Society of America, Volume 93 (5): 2764-2778, May 1993 or S. Spors, R. Rabenstein, and J. Ahrens. The Theory of Wave Field Synthesis Revisited. In proceedings of 124th Convention of the Audio Engineering Society, May 17-20, Amsterdam, The Netherlands, 2008 be removed. It can be used arbitrarily shaped convex or straight speaker arrangements, which need not necessarily be closed. The loudspeaker drive signals can be calculated analytically. In the potential listening area, eg within a loudspeaker arrangement, there is no pronounced "sweet spot" where the reconstruction of the desired sound field is significantly more accurate than in the rest of the listening area. Due to the finite loudspeaker spacing, large deviations from the desired sound field over the entire potential listener area are present in practical implementations (aliasing).

Another family of methods for sound field reconstruction in which the drive signal can be analytically calculated is called Ambisonics. The traditional formulation of Ambisonics (see eg J. Daniel, Representation of champs acoustics, application à la transmission and reproduction of scenes sonores complexes dans un contexte multimedia, PhD thesis, Université Paris 6, 2001 ) requires circular or spherical arrangements of loudspeakers. With the help of numerical algorithms, the loudspeaker signals are generated, which lead to the reproduction of the desired sound field. The limitation of the spatial bandwidth of the drive function necessary in the calculation path has the effect that the reconstruction of the desired sound field in the center of the loudspeaker arrangement is most accurate ("sweet spot"). The further the considered place is from the center, the larger the deviations become.

Extensions to the traditional formulation of Ambisonics (see eg J. Ahrens and S. Spors. Analytical driving functions for higher order Ambisonics. In IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Las Vegas, Nevada, March 30th-April 4th, 2008 ) allow analytical calculation of loudspeaker drive signals that are many times more efficient than numerical methods. However, there is still the restriction that the sweet spot is at the center of the speaker assembly, and that only circular or spherical speaker assemblies can be used.

In addition to these established methods, a number of generalized methods for synthesizing a sound field have been developed in recent years. These are based on the explicit solution of the synthesis equation. This describes the mathematical relationship between the synthesized sound field in a target sound area and the drive signals of a continuous distribution of sound sources through an integral equation. This can be solved by applying operator theory, as in FM Fazi, PA Nelson and R. Potthast, Analogies and Differences between three Methods for Sound Field Reproduction, Ambisonics Symposium, June 2009, Graz, Austria described. For this, the involved sound field quantities are decomposed into orthogonal basis functions. The concrete formulation of these basis functions depends on the underlying geometry. In general, this method is numerically complicated and poorly conditioned. This makes your application complex in practice. However, the formulation of the underlying physical problem in the form of the synthesis equation provides insight into the required mechanisms for the synthesis of a sound field. In the above-mentioned paper by FM Fazi et al. It was shown that the acoustic boundary conditions for the solution of the synthesis equation correspond to those of a scattering body with homogeneous Dirichlet boundary conditions. It can be directly deduced from this that the external sound field of a system for sound field synthesis corresponds to the scattering of the desired sound field at the contour of the system. The boundary conditions at the scattering body then correspond to homogeneous Dirichlet boundary conditions, ie the system for sound field synthesis behaves like a soft-sound scattering body.

Furthermore, in EG Williams, Fourier Acoustics: Sound Radiation and Near Field Acoustic Holography, Academic Press, 1999 a method for the synthesis or extrapolation of a sound field by a continuous distribution of sound sources with monopole characteristic described. Starting from the Kirchhoff-Helmholtz Integral, it is shown that the monopoles must be controlled by the difference of the sound beats from inside and outside in the normal direction to the surface. That is, no explicit solution of the synthesis equation is needed. This was again in the above-mentioned publication by FM Fazi et al. have shown that the underlying requirement for a continuous sound pressure and a discontinuous sound velocity over the edge of the distribution of secondary sources corresponds to the boundary conditions of a soft-sound scattering body. That is, the required difference in the sound velocity in the normal direction can be obtained from the scattering of the virtual sound field at a scattering body with the equivalent geometry as the synthesis system. The outer sound field then corresponds again to the scattering of the virtual sound field at the equivalent scattering body.

Both methods, wavefield synthesis and advanced ambisonics, aim for physically accurate playback in the widest possible audience. In the practical implementation of both methods, the achievable accuracy, however, limits. The finite number of speakers results in a number of artifacts, some of which occur throughout the listening area. This has led to the development of a number of approaches that allow for higher accuracy in a limited audience than wavefield synthesis or ambisonics.

The in the EP-A-2182744 proposed invention allows the free placement of the sweet spot or the sweet area within a closed speaker assembly at Ambisonics. The loudspeaker drive signals can be calculated analytically. The method is significantly more efficient than other approaches, but limited to closed (eg, circular and spherical) arrangements.

The in S. Spors and J. Ahrens. Local sound reproduction by virtual secondary sources, published in AES 40th International Conference on Spatial Audio, Tokyo, Japan, October 2010 The presented approach is based on the concept of spatially dense virtual secondary sources to improve the accuracy of the synthesis. The virtual sources are realized by focused sound sources. This approach can be realized particularly efficiently by the wave field synthesis.

The in J. Ahrens, "The single-layer potential approach applied to sound field synthesis and its extension to non-closing distributions of secondary sources," Ph.D. Dissertation, Technische Universität Berlin, 2010 presented approach is in turn based on a spatial band limitation of the drive signals of the speakers. Thus, a limited range with increased accuracy of synthesis is achieved, which can be placed freely in the handset area. Both enclosing and even loudspeaker arrangements can be used.

The DE-A-10 2007 032 272 and DE-A-10 2005 003 431 describe the technical realization of a virtual headphone. In each case, one or more virtual sound sources are generated in the vicinity of the ears of the listener to simulate a headphone. These virtual sound sources are realized by acoustic focusing. These Virtual sound sources are operated by means of a so-called crosstalk compensation, which compensates for the crosstalk of the signals to the opposite ears out. This is a common technique when playing binaural signals through speakers. The virtual sources are tracked the head movement, thereby the crosstalk compensation can be kept constant during head movements. This is a significant advantage of this invention. The synthesis of outer ear transfer functions to realize other properties than exist in the database is not considered. Furthermore, the sound field synthesis in the method is only used for generating the sources for the virtual headphone and not for the synthesis of a sound field in a local area.

Another method is the so-called time reversal acoustics. The lossless acoustic wave equation contains only temporal and spatial derivatives of second degree (see eg Didier Cassereau and Mathias Fink, Time-Reversal of Ultrasonic Fields Part III: Theory of the Closed-Time Reversal Cavitv, IEEE TRANSACTIONS ON ULTRASONICS, FERROELECTRICS, AND FREQUENCY CONTROL, VOL. 39, NO. SEPTEMBER 1992 ). From this, the interesting property of the wave equation implies that in a known solution of the wave equation, this solution also represents a solution of the wave equation in a reversal of time. Practically, this principle is used to focus sound energy on a point or an object. For this purpose, a sound source is placed at the desired focus point in a first step and the sound pressure is recorded at the later source positions by means of microphones. In a second step, the microphones are replaced by sound sources that are driven by the time-reversed pre-recorded signal. The sound source in the desired focus point and the microphones can also be virtual. The above-mentioned Cassereau and Fink publication mentions applications of time-reversal acoustics, for example, in the destruction of kidney stones by ultrasound.

The present invention is intended to provide an improved and, in particular, a simplified method for reconstructing a sound field in a locally limited listening area.

The present invention is based on the basic idea to combine the findings about the external sound field of a system for sound field synthesis with the time-reversal acoustics for the local synthesis of a sound field. For this purpose it is assumed that the listener area is surrounded by a virtual scattering body. This scattering body can be up to the maximum Range speakers and / or be arbitrarily small. The external sound field is now given by the scattering of the desired virtual sound field at a diffuser whose geometry is equivalent to the geometry of the audience area.

The corresponding acoustic boundary conditions must now be synthesized at the edge of the local listening area. For example, similar to the method of the F.M. Fazi et al. - by an explicit solution of the underlying integral equations. But this requires complex mathematical calculations.

According to the present invention, the principles of time-reversal acoustics are used for this purpose. The field outside the local listener area can be interpreted as the time-reversed field created by a virtual scatterer at the position of the listener area. This external field can be synthesized by applying time reversal through loudspeakers outside the local listener area. This creates the field of the desired virtual source within the local listener area.

Accordingly, the present invention provides a method for determining driving signals for reproducing a desired sound field within a local listening area by means of a plurality of loudspeakers at positions outside the local listener area, comprising the steps of: first, determining a virtual sound field outside the local listener area; Scattering of the desired sound field is created on a virtual diffuser with a geometry that corresponds to the geometry of the local listener area, and the signals of this virtual sound field at the positions of the speakers are determined. The driving signals for the loudspeakers are determined by time-reversing the signals of the virtual sound field at the corresponding loudspeaker positions. For reproducing the desired sound field within the local listener area, after determining the drive signals for the loudspeakers, the loudspeakers to be used are preferably selected from among the plurality of loudspeakers and these are driven by the time-reversed signals.

In the calculation of the scattered field, it is generally assumed that the field within the scattered body corresponds to the incident field. It is further believed that this field can propagate outward from the scatterer. However, these portions of the sound field should preferably not be taken into account in the control of the speakers. Furthermore, the theory of time-reversal acoustics does not take into account the propagation direction of the wave fronts. However, this is crucial for a listener, for example, in the localization of sound sources. Both can be done by a spatial fenestration (selection) in the selection of the used, ie the active speakers.

In the method according to the invention, the loudspeakers are preferably arranged in a circular or linear or rectangular or oval or spherical form.

It is further preferred that the amplitudes of the loudspeaker signals are corrected, as will also be briefly explained below.

The virtual scattering body can have sound-soft or reverberant or mixed boundary conditions.

According to a further preferred embodiment, the virtual sound field is described as generated by a virtual source. Preferably, the virtual source is a point source or a plane wave. In another embodiment, the virtual source is a complex directional source.

More preferably, the method comprises the step of selecting speakers to be used from the plurality of speakers. It can be selected speakers that are not in the propagation direction of the virtual sound field. It is also possible, when selecting the loudspeakers, to select loudspeakers whose propagation direction with respect to the local listener zone coincides with the propagation direction of the virtual sound field.

Preferably, the virtual sound field is determined with the 2.5-dimensional wave field synthesis.

It is further preferred that the listener area be placed dynamically within the speaker assembly. In this case, the listener area can be adapted to the position of a moving listener.

The virtual sound field is preferably a sound field measured by means of microphones outside the audience area.

With the calculation method according to the invention for the drive signals, impulse responses can also be calculated. With these, the signal of the virtual source can then be folded to produce the desired virtual sound field within the local zones. For this purpose, when calculating the sound field which is scattered by the virtual scattering body, it must be assumed that the incident sound field of the virtual source is calculated as a source signal for a temporal (Dirac) pulse. That is, in this case, the space-time impulse response from the virtual scatterer to the loudspeaker positions is then calculated. This must be taken into account when applying the time reversal and possibly also when selecting the loudspeakers.

The method according to the invention can also be applied to measured sound fields. For this purpose, the sound field must be measured with a microphone array, which is located outside the virtual scattering body. From the analysis of the sound field, the incident sound field on the virtual scattering body and thus the scattered field at the loudspeaker positions can be calculated.

The boundary conditions for the virtual scattering body can be set as sound-soft according to the above statements, or alternatively as reverberant.

In a 2.5-dimensional synthesis using wave field synthesis, it may be necessary to correct the amplitudes of the loudspeaker signals. Compared to a three-dimensional wave field synthesis, the 2.5-dimensional synthesis uses secondary sources at the boundary of the (planar) listening area. If point sound sources are used as secondary sources, this is referred to as a 2.5-dimensional synthesis. However, the 2.5-dimensional synthesis suffers from the occurrence of artifacts, in particular amplitude deviations between the desired virtual sources and the synthesized sound field. This is taken into account by correcting the amplitudes of the loudspeaker signals.

Due to the typically small selected listening area and thus small aperture, it is often not useful to use the described method for local sound field synthesis at low frequencies. For low frequencies, the normal methods for sound field synthesis can be used, since the typical physical accuracy of conventional methods is sufficient here. The crossfading between the two methods can be done by a window function in the frequency domain.

In addition to the typical point source model, the virtual source can also be modeled as a source with complex directionality. For this purpose, for example, the in J. Ahrens and S. Spors. Implementation of directional sources in wave field synthesis. In IEEE Workshop on Signal Processing to Audio and Acoustics, New Paltz, USA, October 2007 described method can be used. Again, the combination with the local sound field synthesis according to the invention provides a significant advantage, since the desired directional characteristic is not distorted by the artifacts of spatial sampling.

The inventive method also allows the synthesis of quiet zones within a system for sound field synthesis by the superposition of a synthesized throughout the field sound field and a local sound field. If the sound field within the local listener area has a reversed phase with respect to the global sound field, there is cancellation of the global sound field within the local listener area, and a quiet zone within the local listener area is formed.

In accordance with this aspect, the invention provides a method of creating a quiet zone or zone of reduced sound pressure within a sound field, comprising the steps of: a) determining first drive signals for reproducing the desired sound field within the loudspeaker listening range according to the method of the invention; b) determining second drive signals for reproducing the desired sound field within the area covered by the speakers; c) driving the speakers with the time-reversed signals; and d) phase-reversed superimposition of the drive signals according to step c) with drive signals for the loudspeakers according to step b).

Preferably, step b) is performed by wave field synthesis or near-field compensated higher-order ambisonics.

In principle, the method according to the invention is independent of the technique used for the calculation of the scattered sound field of the virtual scattering body. The scattering of the virtual sound field to a body can, for example, by means of in NA Gumerov and R. Duraiswami, Fast Multipole Methods for the Helmholtz Equation in Three Dimensions, Elsevier, 2004 or by means of numerical methods such as, for example, the 'boundary element method (BEM)' or 'finite element method' (FEM).

The method according to the present invention enables the synthesis of a sound field in a local listener zone, wherein within the local zone with a given number of loudspeakers a higher accuracy is achieved. The desired local playback area can be placed dynamically and freely in front of the loudspeaker arrangement and so e.g. be adapted dynamically to the listener position, in particular by a suitable choice of the position and the geometry of the virtual scattering body. The method can also be used as a substitute for another method of local synthesis in the extrapolation or synthesis of outer ear transfer functions. Furthermore, the generation of quiet zones or zones with reduced sound pressure or reduced energy within the synthesized sound field is possible.

• Fig. 1 zeigt ein Beispiel eines von einem kugelförmigen Streukörper mit schallweichen Randbedingungen gestreuten Felds für eine ebene Welle als einfallendes Schallfeld;
• Fig. 2 zeigt das Schallfeld bei der Synthese einer monofrequenten ebenen Welle mit einer Frequenz (a) von 1 kHz und (b) von 5 kHz mittels traditioneller Wellenfeldsynthese durch eine kreisförmige Lautsprecheranordnung;
• Fig. 3 zeigt die Anwendung des erfindungsgemäßen Verfahrens auf das Szenario aus Fig. 2(b);
• Fig. 4 zeigt das Schallfeld bei der Synthese einer monofrequenten ebenen Welle mit einer Frequenz (a) von 1 kHz und (b) von 4 kHz mittels traditioneller Wellenfeldsynthese durch eine lineare Lautsprecheranordnung;
• Fig. 5 zeigt die Anwendung des beschriebenen Verfahrens zur lokalen Synthese für die Synthese einer ebenen Welle mit einer Frequenz von 4 kHz entsprechend dem in Fig. 4(b) gezeigten Aufbau;
• Fig. 6 zeigt die Ausbildung einer Ruhezone in dem Aufbau gemäß Fig. 2 mittels traditioneller Wellenfeldsynthese und
• Fig. 7 zeigt die Ausbildung einer Ruhezone im Aufbau nach Fig. 6 mittels der vorliegenden Erfindung.

The invention will be described in more detail below with reference to embodiments with reference to the accompanying figures.

• Fig. 1 Fig. 12 shows an example of a plane wave field scattered by a spherical diffuser with soft-sound boundary conditions as an incident sound field;
• Fig. 2 shows the sound field in the synthesis of a monofrequency plane wave with a frequency (a) of 1 kHz and (b) of 5 kHz by means of traditional wavefield synthesis through a circular loudspeaker arrangement;
• Fig. 3 shows the application of the method according to the invention to the scenario Fig. 2 (b) ;
• Fig. 4 shows the sound field in the synthesis of a monofrequency plane wave with a frequency (a) of 1 kHz and (b) of 4 kHz by means of traditional wave field synthesis through a linear loudspeaker arrangement;
• Fig. 5 shows the application of the described method for local synthesis for the synthesis of a plane wave with a frequency of 4 kHz according to the in Fig. 4 (b) shown construction;
• Fig. 6 shows the formation of a quiet zone in the structure according to Fig. 2 using traditional wave field synthesis and
• Fig. 7 shows the formation of a quiet zone in the structure Fig. 6 by means of the present invention.

In a first example, the local synthesis is considered within a 3 meter diameter circular array consisting of 60 loudspeakers. The local listener zone is circular with a diameter of 60 cm and is located in the center of the speaker assembly. FIG. 2 (a) shows the sound field in the synthesis of a monofrequency plane wave with a frequency of 1 kHz by means of traditional wave field synthesis. FIG. 2 (b) shows the same scenario, for a frequency of 5 kHz. Due to the higher frequency and the finite loudspeaker spacing, significant artifacts of the spatial sampling in the synthesized sound field can be observed. FIG. 3 shows the application of the described method to local synthesis. As a virtual scattering body, a sphere with sound-soft boundary conditions was assumed. The ball is also shown in the figures. FIG. 3 clearly shows that at a frequency of the plane wave of 5 kHz an accurate synthesis within the local listener range is possible. Outside the local listening area, artifacts are created.

In the second embodiment, a linear arrangement of a total of 60 loudspeakers is considered, with the distance between the loudspeakers being 15 cm. FIG. 4 (a) shows the synthesis of a monofrequency plane wave with a frequency of 1 kHz. FIG. 4 (b) the same scenario with a frequency of 4 kHz. Again, there are clear artifacts of spatial sampling within the entire listening area. FIG. 5 shows the application of the described method for local synthesis for the synthesis of a plane wave with a frequency of 4 kHz. The improvement in the local listener area, as opposed to FIG. 4 (b) , is clearly visible.

In the third embodiment, the synthesis of a sound field is considered, within which forms a quiet zone. The geometry corresponds to the first example. FIG. 6 shows the synthesis of a monofrequency plane wave with a frequency of 1 kHz using traditional wave field synthesis. FIG. 7 shows the formation of a quiet zone in the center of the loudspeaker arrangement. This was realized by phase-reversed superimposition of the control signals of the traditional wave field synthesis and the proposed local synthesis.

While the invention has been illustrated and described in detail by the figures and the accompanying description, this description and detailed description are to be considered illustrative and exemplary and not limiting as to the invention. It is understood that those skilled in the art can make changes and modifications without departing from the scope of the following claims. In particular, the invention also includes embodiments with any combination of features that are mentioned or shown above in various aspects and / or embodiments.

The invention also includes individual features in the figures, even if they are shown there in connection with other features and / or not mentioned above.

Furthermore, the term "comprising" and derivatives thereof does not exclude other elements or steps. Also, the indefinite article "a" and "derivatives" and derivatives thereof do not exclude a variety. The functions of several features listed in the claims may be fulfilled by one unit. The terms "substantially", "approximately", "approximately" and the like in connection with a property or a value in particular also define precisely the property or exactly the value. All reference signs in the claims are not to be understood as limiting the scope of the claims.

Claims (14)

1. A method for determining control signals for reproducing a desired sound field within a local audience area by means of a plurality of loudspeakers at positions outside the audience area, comprising the steps of:

   - determining a scattered virtual sound field outside the audience area, which is generated by scattering the desired sound field at a virtual scattering body having a geometry that is equivalent to the geometry of the audience area, wherein the virtual scattering body has acoustically soft, acoustically hard or mixed edge conditions,

   - determining the signals of the scattered virtual sound field at the positions of the loudspeakers and

   - determining the control signals for the loudspeakers by time-reversing the signals of the scattered virtual sound field.
2. The method according to claim 1, wherein the loudspeakers are arranged in a circular or linear or rectangular or oval or spherical manner.
3. The method according to claim 1 or 2, wherein the amplitudes of the loudspeaker signals are corrected.
4. The method according to any one of the preceding claims, wherein the scattered virtual sound field is described as being generated by a virtual source.
5. The method according to claim 4, wherein the virtual source is a point source or a plane wave.
6. The method according to any one of the preceding claims, further comprising the step of selecting loudspeakers to be used from the plurality of loudspeakers.
7. The method according to claim 6, wherein loudspeakers are selected which are not located in the propagation direction of the scattered virtual sound field.
8. The method according to claim 6, wherein when selecting loudspeakers, loudspeakers are selected whose propagation direction relative to the local audience zone corresponds to the propagation direction of the scattered virtual sound field.
9. The method according to any one of the preceding claims, wherein the scattered virtual sound field is determined by means of a 2.5-dimensional wave field synthesis.
10. The method according to any one of the preceding claims, wherein the audience area is placed dynamically within the loudspeaker arrangement.
11. The method according to claim 10, wherein the audience area is adapted to the position of a moved auditor.
12. A method for reproducing a desired sound field within a local audience area by means of a plurality of loudspeakers at positions outside the audience area, comprising the steps of:

    - determining control signals for reproducing the desired sound field within the audience area for the loudspeakers in accordance with the method according to any one of claims 1 to 11, and

    - controlling the loudspeakers by means of the time-reversed signals.
13. A method for generating a quiet zone or a zone with reduced sound pressure within a sound field, comprising the steps of:

    (a) determining first control signals for reproducing the desired sound field within the audience area for the loudspeakers in accordance with the method according to any one of claims 1 to 11;

    (b) determining second control signals for reproducing the desired sound field within the area covered by the loudspeakers;

    (c) controlling the loudspeakers by means of the time-reversed signals; and

    (d) phase-reversed superposition of the control signals according to step (c) with the control signals for the loudspeakers according to step (b).
14. The method according to claim 13, wherein step (b) is performed by wave field synthesis or near-field compensated higher-order ambisonics.

Images