Method, apparatus and system for processing audio scene for audio rendering

By sharing and reusing pre-computed diffraction information, the computational complexity problem caused by frequent recalculating of diffraction paths in three-dimensional audio scenarios is solved, and the efficiency and user experience of audio rendering are improved.

CN118985142APending Publication Date: 2024-11-19DOLBY INTERNATIONAL AB
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202380026521.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-10-06
Filing Date
2023-03-02
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

The prior art When using voxels for audio rendering in three-dimensional audio scenarios, frequent recalculation of diffraction paths leads to high computational complexity, increasing device burden and potentially affecting user experience.

Method used

By receiving audio scene descriptions and listener locations, diffraction information is generated and output, and a data interface is provided to share and reuse pre-computed diffraction information, reducing the computational requirement for each rendering instance.

Benefits of technology

Reduces the computing burden of the renderer, improves the efficiency of audio rendering, reduces dependence on powerful computing devices, and improves the user experience.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118985142A_ABST
    Figure CN118985142A_ABST
Patent Text Reader

Abstract

The disclosure relates to a method of processing audio scene information for audio rendering. The method comprises: receiving an audio scene description, the audio scene description comprising a representation of a three-dimensional audio scene and information about a source location of a sound source within the audio scene; receiving an indication of a listener position of a listener within the audio scene; obtaining diffraction information relating to an acoustic diffraction path between a source position and a listener position within the audio scene; performing audio rendering on the sound source based on the diffraction information; and outputting a representation of the diffraction information. The disclosure further relates to a corresponding apparatus, a computer program and a computer readable storage medium.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to U.S. Provisional Application No. 63 / 318,080 filed on March 9, 2022 and U.S. Provisional Application No. 63 / 413,719 filed on October 6, 2022, both of which are incorporated herein by reference in their entirety. Technical Field

[0003] The present disclosure relates to techniques for processing audio scene information for audio rendering. Specifically, the present disclosure relates to voxel-based scene representation and audio rendering. Background Art

[0004] The Moving Picture Experts Group (MPEG) is a consortium of working groups jointly established by the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC) that sets standards for media coding, including audio coding. MPEG is organized under ISO / IEC SC 29, and the audio group is currently identified as Working Group (WG) 6. WG 6 is currently developing a new audio standard (also known as MPEG-I Immersive Audio, ISO / IEC 23090-4).

[0005] The new MPEG-I standard enables auditory experiences from different viewpoints and / or perspectives or listening positions by supporting various motions of a scene and around such a scene, such as motions using various degrees of freedom, such as three degrees of freedom (3DOF) or six degrees of freedom (6DoF), in virtual reality (VR), augmented reality (AR), mixed reality (MR), and / or extended reality (XR) applications. 6DoF interaction extends the 3DoF spherical video / audio experience, which is limited to head rotation (pitch, yaw, and roll), to include translational motion (forward / backward, up / down, and left / right), to enable roaming within a virtual environment (e.g., physically walking around a room) in addition to head rotation.

[0006] For audio rendering in VR, AR, MR, and XR applications, an object-based approach has been widely adopted by representing complex auditory scenes as multiple individual audio objects, each associated with parameters or metadata that define the object's orientation / position and trajectory in the scene. Alternatively, audio rendering in such environments also uses high-order ambisonics (HOA). However, new uses of "voxels" for rendering audio scenes are now being explored, such as for new immersive audio experiences. Voxels used for audio rendering are related to media environments (such as video games and / or VR, AR, MR, and XR environments) implemented in both hardware and software.

[0007] A voxel is a volume of space to which acoustic properties or audio rendering instructions are assigned. The voxel size may be an encoder configuration parameter, and it may be selected (manually or automatically) according to the geometric detail level of the scene (e.g., in the range of 10cm-1m).

[0008] Voxels for audio rendering can be obtained by:

[0009] Voxelize (or convert) a grid-based scene representation

[0010] Based on the scene representation used for scene generation (or even video rendering) (e.g. by downsampling to a smaller size voxel)

[0011] However, traditional methods for providing realistic sound for user experiences (including those involving motion) in VR, AR, MR, and XR environments using voxels remain challenging and computationally complex.

[0012] Typical techniques for diffraction modeling in three-dimensional audio scenes (such as for computer-mediated reality applications) require recalculation of diffraction paths and other diffraction information when any of the audio scene, user position, or audio source position changes. For example, diffraction paths may change when a user and / or an audio source moves through the three-dimensional audio scene. Further, diffraction paths may change when the audio scene itself changes, such as by indicating the opening or closing of a door or window, etc. Frequent recalculation of diffraction paths may be computationally expensive, requiring relatively powerful computing devices to implement computer-mediated reality applications and / or may negatively impact the user experience in some cases.

[0013] Therefore, there is a need for improved techniques for diffraction modeling in three-dimensional audio scenes, particularly three-dimensional audio scenes utilizing voxels. Such techniques are particularly desirable so that the computational burden of devices (e.g., decoders, renderers) implementing such techniques can be reduced. Summary of the invention

[0014] In view of this need, the present disclosure provides a method for processing audio scene information (particularly voxel-based audio scene information) for audio rendering, an apparatus for processing audio scene information for audio rendering, a computer program and a computer-readable storage medium, each of which has the features of the corresponding independent claims.

[0015] One aspect of the present disclosure relates to a method for processing audio scene information for audio rendering. The method may include receiving an audio scene description. The audio scene description may include a representation of a three-dimensional audio scene and information about source positions of sound sources within the audio scene. The method may further include receiving an indication of a listener position of a listener within the audio scene. The method may further include obtaining diffraction information related to an acoustic diffraction path between a source position within the audio scene and the listener position. The method may further include performing audio rendering on the sound source based on the diffraction information. The method may further include outputting a representation of the diffraction information. The output of the representation of the diffraction information may reach at least one external (e.g., shared) data source or repository so that the diffraction information can be reused by an external rendering or decoder instance.

[0016] In addition to local reuse, the proposed method also provides an interface (e.g., a data interface for exchanging data (including, for example, a predefined format of representations of diffraction information)) for sharing generated / computed diffraction information between rendering instances (or decoder instances). The interface can be implemented in a software format that can be executed on one or more hardware platforms. For example, the interface can be a graphical user interface that displays information and / or allows user interaction. This can be used to establish a framework for rendering instances between which locally generated (or even externally obtained) diffraction information is shared. Thereby, the computing power required for each rendering instance to render can be reduced. Importantly, the diffraction information is generated on the decoder side and is therefore automatically ensured to be applicable to real-life use cases and situations. In particular, a large amount of diffraction information will be available for listener positions in the audio scene that are frequently visited by actual users. Accordingly, the amount of storage used to store representations of the diffraction information (e.g., shared / physical storage or bitstream) can be used efficiently and will be used to store representations of the actually relevant diffraction information to a large extent.

[0017] In some embodiments, outputting a representation of the diffraction information may include outputting a data element including the diffraction information and information about a scene state. The scene state may include an audio scene description and a listener position. The output of the data element may go to a bitstream or a storage device. In particular, the output may go to at least one external (e.g., non-local, in particular shared) data source or repository (e.g., shared / cloud storage or a bitstream) such that the diffraction information can be reused by an external rendering instance.

[0018] The diffraction information may be reused by the same device / decoder / renderer at a later point in time, or may be used by other devices / decoders / renderers. A device / decoder / renderer may use the scene state included in the data element to determine whether the available diffraction information is applicable for a given configuration of the audio scene and the listeners therein, or in other words, whether the diffraction information is available for a given configuration.

[0019] In some embodiments, the representation of the diffraction information may be output to a bitstream (outgoing bitstream) and / or to a storage device.

[0020] In some embodiments, the diffraction information may be output for later reuse in audio rendering by the same rendering instance, or for later reuse by another rendering instance.

[0021] In some embodiments, in accordance with the MPEG-I standard or any subsequent version of the MPEG-I standard, a representation of the diffraction information may be output as part of a voxSceneDiffractionPreComputedPathData() syntax element.

[0022] In some embodiments, the diffraction information may indicate a virtual source position of a virtual sound source. The virtual sound source may be selected to "encapsulate" the application of diffraction and / or occlusion effects to the sound source so that the virtual sound source sounds the same or substantially the same when rendered directly as when rendered through diffraction and / or occlusion processing. When viewed from the listener position, the virtual source position may have the same direction (e.g., azimuth, or azimuth and elevation) as a first position (diffraction angle) on or near the sound diffraction path, for which the diffraction path changes direction and a straight line from the diffraction angle voxel to the listener voxel position is not occluded. The virtual source distance may correspond to the length of the diffraction path. The diffraction information may include the following defined C vox and r in Instructions, where, in short, C vox indicates the position (e.g., voxel position) of the diffraction angle and r in Indicates the length of the diffraction path.

[0023] In some embodiments, the representation of the three-dimensional audio scene may be a voxel-based representation. The representation of the three-dimensional audio scene may then include one or more indications of a cubic volume in a voxel grid, and wherein each such indication may include information about a pair of polar voxels defining the cube and information about common voxel properties of the voxels in the cubic volume. This makes the voxel-based representation of the three-dimensional audio scene more efficient.

[0024] In some embodiments, information about a pair of polar voxels of a cubic volume may include an indication of the respective voxel indices assigned to the polar voxels. Here, voxels of a voxel-based audio scene representation may have uniquely assigned consecutive voxel indices. This makes the representation of voxel coordinates more efficient.

[0025] In some embodiments, the representation of the three-dimensional audio scene may be a voxel-based representation. The diffraction information may then include an indication of the position of a voxel located on or near a diffraction path and an indication of the length of the diffraction path. Further, the indication of the position of a voxel located on or near the diffraction path may be an indication of a voxel index assigned to the voxel, wherein the voxels of the voxel-based audio scene representation may have uniquely assigned consecutive voxel indices.

[0026] In some embodiments, the method may further comprise determining a (current) scene state based on the audio scene description and the listener position. The current scene state may then be used to determine whether pre-computed diffraction information is available for the current configuration of the audio scene and the current listener position.

[0027] In some embodiments, the method may further include determining whether the current scene state corresponds to a known scene state for which pre-computed diffraction information may be obtained. For example, the pre-computed diffraction information may be obtained from a bitstream (incoming bitstream) or a storage device (particularly including an external storage device, such as a shared / cloud storage device).

[0028] In some embodiments, determining whether the current scene state corresponds to a known scene state may include determining a hash value based on the current scene state. This may further include comparing the determined hash value with a hash value of the known scene state.

[0029] In some embodiments, the method may further include determining the diffraction information by extracting pre-computed diffraction information of the known scene state from a bitstream or a storage device (e.g., a local memory, a cache, an external memory, a shared memory, a cloud-implemented memory, etc.) if it is determined that the current scene state corresponds to a known scene state. In particular, the pre-computed diffraction information may be obtained at least in part from an external data source or repository.

[0030] In some embodiments, the method may further comprise determining diffraction information using a pathfinding algorithm based on the source positions, the listener positions and the representation of the three-dimensional audio scene if it is determined that the current scene state does not correspond to a known scene state.

[0031] In some embodiments, the method may further include receiving a lookup table or an entry of the lookup table from a bitstream or a storage device, the lookup table comprising a plurality of pre-computed diffraction information, each pre-computed diffraction information being associated with a corresponding known scene state. Thus, the LUT may involve or include a plurality of the aforementioned data items. The known scene state may include a known audio scene description and a known listener position.

[0032] In some embodiments, the representation of the three-dimensional audio scene may be a voxel-based representation.

[0033] According to another aspect, a method of compressing an audio scene for three-dimensional audio rendering is provided. The method may include obtaining a voxelized representation of the audio scene, the voxelized representation including a plurality of voxels arranged in a voxel grid, each voxel having associated voxel attributes. The method may further include determining, among the voxels of the voxelized representation, a set of voxels forming a connected geometric region on the voxel grid, wherein the voxels in the geometric region share common voxel attributes. The method may further include generating a representation of the audio scene based on the determined set of voxels.

[0034] In some embodiments, the geometric region may have a cubic shape. Then, the method may further include determining at least a first boundary voxel and a second boundary voxel of the set of voxels from the plurality of voxels of the voxelized representation. Wherein, the first boundary voxel and the second boundary voxel may define the cubic shape of the geometric region.

[0035] In some embodiments, the voxel attributes of each voxel may include acoustic properties associated with the voxel and / or a set of audio rendering instructions assigned to the voxel. Further, the common voxel attributes of the voxels in the geometric region may include common acoustic properties associated with those voxels and / or a set of common audio rendering instructions assigned to those voxels.

[0036] In some embodiments, the method may further include determining at least one scene element parameter for the geometric region, the at least one scene element parameter comprising one or more of the following: a scene element identifier, an acoustic property identifier, and / or an audio rendering instruction set identifier, and indices of corresponding first and second bounding voxels defining the geometric region. Here, a scene element is understood to be associated with a geometric region and its corresponding voxel properties (e.g., material properties and / or rendering instructions) and optionally an identifier (e.g., a scene element identifier).

[0037] In some embodiments, the method may further comprise applying entropy encoding and / or lossy encoding to at least one scene element parameter of the geometric region in a sequential (processing all data) or progressive (processing part of the data) manner.

[0038] In some embodiments, the method may further comprise outputting a bitstream comprising at least one scene element parameter to determine a set of voxels associated with a geometric region of the compressed representation of the audio scene based on the determined set of voxels.

[0039] In some embodiments, the geometric regions may be associated with scene elements within the audio scene.

[0040] In some embodiments, the audio scene may include a large scene represented by the determined set of voxels. The large scene may include a set of sub-scenes. Each sub-scene may correspond to a subset of the determined set of voxels. Then, the method may further include determining a subset of voxels corresponding to the sub-scene in the determined set of voxels.

[0041] In some embodiments, the method may further include applying interpolation and / or applying filtering to the audio properties of the voxels and the renderer instructions in time and / or space.

[0042] In some embodiments, the method may further include redefining voxel properties of a subset of a set of voxels associated with a scene sub-element in the geometric region for overwriting the subset with the redefined voxel properties. Here, a scene sub-element is understood to be associated with a scene element whose geometric region is included (e.g., completely included) within the geometric region of another scene element.

[0043] In some embodiments, the method may further include determining a voxel superset that includes the determined set of voxels. The determined set of voxels may be associated with a scene sub-element within the geometric region. The method may then further include assigning new voxel attributes to the determined set of voxels and overwriting the voxel attributes of the determined set of voxels with the new voxel attributes.

[0044] In some embodiments, the method may further include determining a voxel size for representing the geometric region, wherein the voxel size is based on a number of voxels along a scene dimension.

[0045] According to another aspect, a device for processing audio scene information for audio rendering is provided. The device may include a processor and a memory, the memory being coupled to the processor and storing instructions for the processor. The processor may be configured to perform all steps of the method according to the aforementioned aspect and its embodiments.

[0046] According to a further aspect, a computer program is described. The computer program may include executable instructions for performing the method or method steps outlined throughout the present disclosure when executed by a computing device (eg, a processor).

[0047] According to another aspect, a computer-readable storage medium is described. The storage medium may store a computer program adapted to be executed on a computing device (eg, a processor) and for performing the method or method steps outlined throughout the present disclosure when executed on the computing device.

[0048] It should be noted that the methods and systems as outlined in this disclosure (including preferred embodiments thereof) can be used independently or in combination with other methods and systems disclosed in this document. In addition, all aspects of the methods and systems outlined in this disclosure can be combined arbitrarily. In particular, the features of the claims can be combined with each other in any manner.

[0049] It will be understood that device features and method steps can be interchanged in a variety of ways. In particular, as will be appreciated by the skilled person, details of the disclosed method(s) may be implemented by corresponding devices, and vice versa. Furthermore, any statements above regarding method(s) (and, for example, steps thereof) should be understood to apply equally to corresponding devices (and, for example, blocks, stages, units thereof), and vice versa. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The present invention is explained below in an exemplary manner with reference to the accompanying drawings, in which:

[0051] Figure 1 schematically illustrates an example of a processing chain for processing audio scene information for audio rendering;

[0052] Figure 2 schematically illustrates examples of diffraction paths for source and listener positions in a voxel-based three-dimensional audio scene;

[0053] Figure 3 is a flowchart illustrating an example of a method of processing audio scene information for audio rendering according to an embodiment of the present disclosure;

[0054] Figure 4 is a diagram illustrating an embodiment according to the present disclosure Figure 3 Flowcharts of examples of implementation details of the method;

[0055] Figures 5 to 7 schematically illustrates an example of a processing chain for processing audio scene information for audio rendering according to an embodiment of the present disclosure;

[0056] Figure 8 is a graph illustrating a complexity metric variation over time for different operation modes / implementations of processing audio scene information for audio rendering according to an embodiment of the present disclosure;

[0057] Fig. 9schematically illustrates examples of possible use cases for techniques according to embodiments of the present disclosure;

[0058] FIG. 10A to FIG. 10C schematically illustrates an example of a portion of a voxel-based audio scene according to an embodiment of the present disclosure;

[0059] Fig.11 schematically illustrates an example of a voxel-based audio scene to which embodiments of the present disclosure may be applied; and

[0060] Fig.12 is a block diagram schematically illustrating an example of an apparatus for implementing a method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0061] Hereinafter, exemplary embodiments of the present disclosure will be described with reference to the accompanying drawings. The same elements in the drawings may be indicated by the same reference numerals, and repeated descriptions thereof may be omitted.

[0062] Voxel-based audio scene representation

[0063] First, an overview of voxel-related concepts of audio scene representation will be given.

[0064] What are voxels for audio rendering?

[0065] A voxel is understood to be a volume of space to which acoustic properties or audio rendering instructions are assigned.

[0066] What is the voxel size for audio rendering?

[0067] The voxel size may be an encoder configuration parameter. It may be selected (manually or automatically) based on the geometric detail level of the scene (e.g., in the range of 10cm-1m).

[0068] How to handle large audio scenes?

[0069] Large audio scenes do not necessarily result in a large number of voxels and high rendering complexity. For example, a large audio scene can be represented as

[0070] A set of independent sub-scenes (and a way to "teleport" between these representations without requiring a renderer "restart")

[0071] A set of scene updates (based on user location)

[0072] How to deal with discontinuities caused by voxel granularity?

[0073] Any strong discontinuities in the sound level (and jumps in the direction of the diffracted signal) may be avoided by applying interpolation (eg in time and space).

[0074] How to represent a voxel-based audio scene?

[0075] Any voxel-based audio scene representation may include an indication of voxels that are not transmission voxels (e.g., are occlusion voxels) (i.e., voxels in which sound cannot propagate or cannot propagate freely) - a representation of the occlusion geometry. The indication may involve an indication of the coordinates of the corresponding voxels (e.g., center coordinates, angular coordinates, etc.). For example, the coordinates of these voxels may be represented by a grid index. In addition, the voxel-based representation may include an indication of material properties (such as absorption coefficients, reflection coefficients, etc.) of voxels that are not transmission voxels. In addition to occlusion voxels, the voxel-based representation may also indicate transmission voxels (e.g., air voxels), i.e., voxels in which sound can propagate - a representation of the sound propagation medium. Accordingly, for each voxel in a predefined portion of space (e.g., within a boundary surrounding the audio scene), some embodiments of the voxel-based audio scene representation may include an indication of the corresponding material properties.

[0076] Techniques for processing audio scene information

[0077] Figure 1 A processing chain 100 is schematically illustrated, which can be used to process audio scene information for audio rendering. Specifically, the processing chain 100 can be used to convert voxel-related data into parameters and signals required for sonification (or general audio rendering). The processing chain 100 can be implemented in software, hardware, or a combination thereof. For example, the processing chain 100 can be implemented by a renderer / decoder coupled to an AR / VR / MR / XR device (such as AR / VR / MR / XR goggles). Specific embodiments may include game consoles, set-top boxes, personal computers, etc.

[0078] The processing chain receives an audio scene description 20 from a bitstream (or storage / memory) 10. The audio scene description 20 may include a representation of a three-dimensional audio scene and information about source positions of sound sources within the audio scene. For example, the representation of the three-dimensional audio scene may be voxel-based.

[0079] The processing chain 100 further receives an indication of a user position (listener position) 30 of a user (listener) within the audio scene. The audio scene description 20 and the user position 30 are provided to a diffraction direction calculation block (diffraction calculation block) 40 to determine (e.g., calculate) diffraction information. The diffraction information may be related to the acoustic diffraction path between the source position and the listener position within the audio scene. The diffraction information is then provided to a diffraction modeling tool 50 to apply diffraction modeling and optionally occlusion modeling based on the diffraction information. The occlusion modeling calculates the attenuation gain of a straight line between the listener and the audio source. The diffraction modeling tool 50 may output sonified audio data (3DoF sonifier data), the sonified audio data including, for example, the position, orientation, and frequency-dependent gain of the object to be rendered. The diffraction modeling tool output may be further processed by other rendering stages (e.g., Doppler, directivity, distance attenuation, etc.). In general, the diffraction modeling tool 50 may be said to output diffraction information, as described in detail below. For example, the sonified audio data may then be used for audio playback.

[0080] In short, if Figure 1 The processing chain shown may be used to convert voxel related data into parameters of sonification and signals. The diffraction direction calculation block 40 and the diffraction modeling tool 50 may be considered as non-limiting examples of rendering tools. In general, the rendering tool may generate 3DoF sonifier data.

[0081] As described above, a scene description may include a voxel matrix and associated coefficients (e.g., reflection coefficients, occlusion coefficients, absorption coefficients, transmission coefficients, etc.). These coefficients may indicate the material or material properties of the corresponding voxels. Rendering tools may include, for example, occlusion and diffraction modeling tools. 3DoF sonifier data may include, for example, object positions, orientations, and frequency-dependent gains.

[0082] As described above, a voxel-based representation of a three-dimensional audio scene defines psychoacoustically relevant geometric elements and sound propagation media. In some embodiments, the scene description can provide information to the rendering tool using the following parameters / interfaces (e.g., the following agreed data formats, or agreed data exchange points):

[0083] Scene size:

[0084] - expressed in absolute units (e.g., meters)

[0085] - expressed as number of voxels and / or voxel size

[0086] Scene Anchor:

[0087] - In terms of coordinate anchors (mapping absolute coordinates to voxel indices)

[0088] - In terms of scene anchors (mapping a subscene to a subset of voxels)

[0089] Scene content data:

[0090] - refers to material properties (e.g., transmission coefficient, reflection coefficient, etc.) that approximate the acoustic effects caused by occluders (sound obstacles) located in the corresponding volume

[0091] - refers to the properties of the sound propagation medium that approximate the acoustic effects caused by the medium located in the corresponding volume (e.g., sound speed, energy absorption, distance decay curve, etc.)

[0092] - Rendering control parameters that describe the desired occlusion modeling effect

[0093] For example, a "global" or "local" occluder type that determines the length (and shape) of the occlusion effect shadow behind that voxel.

[0094] - Rendering control parameters describing the expected sound diffraction modeling effect

[0095] For example, voxel types that control / cause sound directionality changes (i.e. diffracted sound paths cannot penetrate the volume) - content control parameters that describe audio signal correlation and scene creation

[0096] For example, determining which signal has perceptual relevance (rendering) in the corresponding volume, the audio signal ID and / or signal gain

[0097] - Rendering control parameters describing the desired reverberation modeling effect

[0098] For example, voxel type that controls reverb settings (e.g., RT60, DDR, RIR, etc.)

[0099] Scene content update:

[0100] - refers to the update trigger event

[0101] All data can be audio object related (to support the content creator's intent in flexible audio scene authoring).

[0102] 3DoF auralizer data may include the following information:

[0103] - A set of audio object (and HOA) parameters and associated signals

[0104] ○ Parameters include metadata output from rendering tools (i.e. position, orientation and gain for simulated occlusion, diffraction, early reflection effects, parameters for reverberation coefficients, IR, etc.)

[0105] o The associated signal represents the audio output of the rendering tool (i.e. the downmixed or copied audio signal)

[0106] - Scene state identifier (i.e. metadata that allows mapping of scene description and user input to 3DoF sonifier data)

[0107] Figure 2 Examples of possible scene states and diffraction paths for the scene states are illustrated.It should be understood that a scene state relates to or comprises a listener position 210 and an audio scene description (comprising a representation of a three-dimensional audio scene and source positions 220).

[0108] Figure 2 An example relates to a voxel-based representation of a three-dimensional audio scene. This voxel-based representation indicates "air" voxels or empty voxels (i.e., voxels in which sound can propagate or transmission voxels) 230 and occluded voxels 240 (i.e., voxels in which sound cannot propagate or cannot propagate freely). Accordingly, occluded voxels can be understood to relate to voxels that are filled with materials other than air and can reflect, block or otherwise change sound propagation. For the occluded voxels 240, the representation can further indicate corresponding transmission coefficients, reflection coefficients, and potential absorption coefficients related to the material properties of these voxels. These coefficients can be related to the ID or index of their corresponding voxels in the voxel-based representation. In general, the voxel-based representation can define psychoacoustically relevant geometric elements and sound propagation media in an audio scene.

[0109] The listener position 210 is represented by the parameter L VOX Indicates, and the source position 220 is indicated by another parameter S VOX instruct.

[0110] A diffraction path between the source position 220 and the listener position 210 may be determined using a pathfinding algorithm that takes as input the listener position 210, the source position 220, and a representation of the three-dimensional audio scene (or a two-dimensional representation derived therefrom, such as a 2D projection or a 2D matrix). For example, an algorithm for determining diffraction information may take as input the listener position 210, the source position 220, and a representation of the three-dimensional audio scene, and may output a pathfinding algorithm using C VOX The position of the diffraction angle 250 indicated and the variable r representing the diffraction path length in For example, diffraction information can be determined based on:

[0111] [C vox ,r in ]=DiffractionDirectionCalculation(L vox ,S vox ,VoxDataDiffractionMap)

[0112] Where DiffractionDirectionCalculation indicates the algorithm used to determine the diffraction information ("pathfinding algorithm"), and VoxDataDiffractionMap indicates a voxel-based representation of the three-dimensional audio scene or a processed version thereof (e.g., a 2D projection or a 2D matrix derived therefrom). VOX It is understood as a coordinate indicating a diffraction angle (eg a coordinate of a corresponding voxel comprising the diffraction angle, a voxel / grid coordinate or a voxel / grid index).

[0113] Here, DiffractionDirectionCalculation may involve any feasible pathfinding algorithm, for example, a fast traversal algorithm for ray tracing (reference Amanatides, J. and A.Woo, A Fast Voxel TraversalAlgorithm for Ray Tracing. Proceedings of EuroGraphics, 1987.87.) and a JPS algorithm (reference Harabor, DD and A.Grastien, Online Graph Pruning for Pathfinding On GridMaps. Proceedings of the Twenty-Fifth AAAI Conference on ArtificialIntelligence, 2011.). Further, a 3D path search algorithm may be directly applied using a voxel-based scene representation to obtain the shortest path between the source position 220 and the listener position 210. Alternatively, a 2D path search algorithm may be applied for this task using an appropriate 2D projection plane of the 3D voxel-based scene representation. For indoor (e.g., multi-room) sound simulation, the corresponding 2D projection plane may be similar to a plan view describing a "sound propagation path topology". For outdoor sound simulation scenarios, it may be interesting to consider a second (e.g., vertical) 2D projection plane to account for diffraction paths across sound obstacle(s) or occluding structure(s). The pathfinding method remains the same for all projection planes, but its application provides additional paths that can be used for diffraction modeling.

[0114] Assume that the pathfinding algorithm outputs a diffraction path connecting the source position 220 with the listener position and consisting of a plurality of straight path segments (line segments) linked sequentially end-to-end. Each transition from one path segment to another is associated with a change in direction of the diffraction path.

[0115] According to the algorithm used to determine the diffraction information, the diffraction angle C VOXThe angular voxels (a set of voxels C representing the angular voxels on the diffraction pattern) that are located on or near the diffraction path and that correspond to the diffraction pattern (indicated by the voxel-based representation) can be determined. set For example, the diffraction angle C VOX The diffraction path can be obtained from a set of voxels (P set ) is selected as the approach leading to the path (P set ) changes direction (from the listener position Lc) of the "visible" corner voxel (belonging to C set ). If there is more than one such angle, the voxel along the diffraction path (P set )The corner farthest from the listener position.

[0116] In general, a diffraction path algorithm can be said to determine diffraction information related to an acoustic diffraction path between a source position and a listener position within an audio scene.

[0117] The diffraction information may be sufficient information for a renderer to recover / determine a virtual source position of a virtual audio source that encapsulates the effect of the acoustic diffraction effect. VOX The coordinates and diffraction path length r in This is the case. For example, when viewed from the listener position, the virtual source position can be recovered by calculating the direction of the diffraction angle (e.g., azimuth, or azimuth and elevation). Using this direction and adding the path length r of the diffraction path in As a virtual source distance to the listener position, a virtual source position may be determined.

[0118] Note that the diffraction information can be represented in different ways. As mentioned above, one option is to include / store the path length r in and the diffraction angle C VOX Diffraction information of coordinates (e.g., grid coordinates, etc.).

[0119] Based on the above, the following data elements can be defined:

[0120] An example of scene state N1 can be expressed as

[0121] N1={L vox ,S vox ,VoxDataDiffractionMap},

[0122] That is, it may involve or include the listener position L of the audio scene. VOX , source location S VOX and voxel-based representations (e.g., VoxDataDiffractionMap).

[0123] The scene state identifier of scene state N1 can be defined as

[0124] SceneStateIdentifier=HASH(N1),

[0125] HASH is a hash function that generates a hash value for the scene state N1, for example, it maps the scene state to a value of a fixed size. Generally speaking, the scene state identifier can be said to indicate a scene state or identify a scene state.

[0126] Further, an example of diffraction information N2 can be expressed as

[0127] N2={C vox ,r in},

[0128] As mentioned above, r in is the path length of the diffraction path, and C vox Indicates the location (eg, voxel position) of the diffraction angle.

[0129] The quantized version of the diffraction information N2 can be indicated by N3, where

[0130] N3=voxSceneDiffractionPreComputedPathData(N1),

[0131] N3(~=N2)=DiffractionDirectionCalculation(N1),

[0132] voxSceneDiffractionPreComputedPathData() parses the bitstream and retrieves the precomputed (stored and quantized) diffraction information (e.g., Figure 6 The bitstream syntax of the processing chain 500 is generated), and DiffractionDirectionCalculation() represents a function that performs online calculation of diffraction information, which can be used, for example, in Figure 5 and Figure 7 The diffraction direction calculation block 40 is implemented in the above method.

[0133] Diffraction information (e.g. C vox and r in ) can also be considered to be related to the 3DOF sonifier data, because the user position voxel coordinates L vox is fixed.

[0134] Table 1 gives an example of the syntax element voxSceneDiffractionPreComputedPathData() according to the MPEG-I standard. The voxel payload data structure may have the following elements:

[0135] numberOfVoxDiffractionPathData

[0136] This element indicates the number of precomputed diffraction path data sets.

[0137] voxDiffractionPathStartVoxelPacked

[0138] This element represents the packed form of the variable voxDiffractionPathStartVoxel, which indicates the voxel index of the path start voxel of the pre-computed diffraction path (e.g., Figure 2 S vox or L vox ). For example, voxDiffractionPathStartVoxel can be a 2D position on the diffraction pattern indicating the starting position of the diffraction path.

[0139] voxDiffractionPathEndVoxelPacked

[0140] This element represents the packed form of the variable voxDiffractionPathEndVoxel, which indicates the voxel index of the path end voxel of the precomputed diffraction path (e.g., Figure 2 L vox or S vox ). For example, voxDiffractionPathEndVoxel can be a 2D position on the diffraction pattern indicating the end position of the diffraction path.

[0141] voxDiffractionPathDataExistFlag

[0142] This element indicates whether a diffraction path exists.

[0143] voxDiffractionSourceDirectionPacked

[0144] This element represents a packed form of the variable voxDiffractionSourceDirection, which indicates the voxel index of the voxel used to determine the diffraction source azimuth value. For example, this may correspond to the angle voxel C vox .

[0145] voxDiffractionPathLength

[0146] This element represents diffraction Figure 2The diffraction path length on the matrix D. For example, this can correspond to the path length r in .

[0147] Note:

[0148] voxSceneDimensions

[0149] The number of voxels per scene dimension.

[0150] escapedValue()

[0151] This element implements a generic method for transmitting integer values ​​using varying numbers of bits. It features a two-level escape mechanism that allows the range of representable values ​​to be extended by transmitting additional bits in succession. The syntax of escapedValue() shall be as defined in ISO / IEC 23003-3.

[0152] Table 1—voxSceneDiffractionPreComputedPathData() syntax

[0153]

[0154] Variables obtained from voxSceneDiffractionPreComputedPathData() (e.g. Figure 6 As shown) is further processed to output diffraction information N3.

[0155] A technical benefit and effect of the techniques according to the present disclosure is that if the scene state has been processed accordingly and the diffraction information or 3DoF sonifier data is available, the scene state identifier or other information derived from the scene state can be used to avoid applying a diffraction modeling tool or a rendering tool. In this case, the renderer can access the diffraction information / 3DoF sonifier data (for a known scene state) without applying a rendering tool by:

[0156] - reuse previously calculated (precomputed) data, or

[0157] -Applies data calculated by another renderer.

[0158] Therefore, a technical benefit and effect is that the technology according to the present disclosure relates to lossless functionality for low complexity modes (complexity relative to bit rate).

[0159] In order to fully implement such a scheme, the present disclosure proposes to provide (e.g. in a decoder / renderer) a processing chain for processing audio scene information for audio rendering with an interface for providing / outputting diffraction information for later use or for use by a different decoder / renderer. The interface is understood as a data interface for outputting data in a predefined format to allow continuous reuse (especially for reuse by other decoders / renderers). The interface can be implemented and / or utilized in any combination of software and hardware. In particular, this may involve providing / outputting data elements including diffraction information and information about a scene state (such as a scene state identifier). The data element may have a predefined format, for example with predefined data fields. Using the interface, the processing chain may provide the calculated diffraction information or 3DoF auralizer data together with the scene state identifier to other decoders / renderers and / or store it for later reuse.

[0160] Example 1: If the decoder / renderer has obtained diffraction information (e.g., diffraction path) for a given user position (listener position), the decoder / renderer can reuse the diffraction information until the user leaves the corresponding voxel volume (or the scene description is updated).

[0161] Example 2: If the computed diffraction information corresponds to a scene state that is unknown to other decoders, then those decoders can reuse the diffraction information and avoid running their own diffraction modeling tools or rendering tools.

[0162] The exchange and sharing of diffraction information between different decoders may be accomplished using a database, which may be included in the bitstream (eg, accessed via application request).

[0163] Figure 3 is a flow chart illustrating an example of a method 300 for processing audio scene information for audio rendering according to an embodiment of the present disclosure. The method 300 may be implemented in software, hardware, or a combination thereof. For example, the processing chain 100 may be implemented by a renderer / decoder coupled to an AR / VR / MR / XR device such as AR / VR / MR / XR goggles. Specific implementations may include game consoles, set-top boxes, personal computers, and the like.

[0164] The method 300 comprises steps S310 to S350, which may be performed, for example, by a decoder / renderer. For example, these steps may be performed whenever a scene state changes. Where the scene state is understood to involve or include a listener position 210 and an audio scene description (including a representation of a three-dimensional audio scene and source positions 220), such as implemented by the above scene state N1, a change in the scene state may involve one or more of a change in the listener position 210, a change in the source positions, and a change in (the representation of) the three-dimensional audio scene. Alternatively, steps S310 to S350 may be performed for each of a plurality of processing cycles of the decoder / renderer. However, if the audio scene description does not change, step S310 may be omitted. It should also be understood that steps S310 to S350 need not be repeated in order to be repeated. Figure 3 Execute in the order shown.

[0165] exist Step S310 At the audio scene description, an audio scene description is received. The audio scene description includes a representation of a three-dimensional audio scene and information about source positions of sound sources within the audio scene. For example, the audio scene description may include elements such as those defined above. vox and VoxDataDiffractionMap.

[0166] exist Step S320 At , listener position information of a listener in the audio scene is received. For example, the listener position may correspond to the element L defined above. VOX .

[0167] exist Step S330 At , diffraction information related to the acoustic diffraction path between the source position and the listener position in the audio scene is obtained. The obtained diffraction information may indicate a virtual source position of the virtual sound source. For example, when viewed from the listener position, the virtual source position may have a diffraction angle C VOX The virtual source distance may correspond to the length r of the diffraction path. in Accordingly, the diffraction information may include the C defined above. vox and r in instructions.

[0168] exist Step S340 At the same time, audio rendering is performed on the sound source based on the diffraction information. This may include, for example, diffraction modeling.

[0169] To this end, a virtual source position of a virtual source may be determined based on the diffraction information. The virtual source may be an audio source that encapsulates the acoustic diffraction effect between the source position and the listener position in the three-dimensional audio scene. For example, the virtual source position may be based on C VOX and r in This is determined by:

[0170] Determine the diffraction angle C when viewed from the listener position VOX direction (e.g., azimuth or azimuth and elevation);

[0171] using the determined direction as a virtual source direction of the virtual source when viewed from the listener position; and

[0172] Use the diffraction path length r in As

[0173] ○ the virtual source distance from the listener position, or

[0174] ○ Virtual source gain compensation from diffraction and direct path length.

[0175] For example, audio rendering may then include rendering virtual sound sources at the virtual source locations.

[0176] exist Step S350 At the same time, outputting a representation of the diffraction information. For example, outputting a representation of the diffraction information may include outputting a data element including the diffraction information and information about a scene state. The scene state may include an audio scene description (e.g., S vox and VoxDataDiffractionMap) and listener position (e.g., L VOX ).

[0177] The output may be provided to a look-up table (LUT). The LUT comprises as its entries different diffraction information items indexed with information about the respective scene state (e.g. indexed with the respective scene state identifier). Thus, the LUT may be said to comprise diffraction information and information about the scene state. The LUT may be stored and / or provided for later retrieval, e.g. by other decoders, e.g. by application request, from the bitstream or from a shared storage device (e.g. cloud or server based). The actual desired entry may be retrieved from the LUT using a hash value of the scene state or a scene state identifier.

[0178] Further, the representation of the diffraction information can be output to a bitstream (e.g., an outgoing bitstream) and / or a storage device (e.g., a memory, a cache, a file, etc.). The storage device can be local, or can be shared (e.g., cloud-based). In general, the representation of the diffraction information can be output to a suitable medium for storing digital information or computer-related information. The output can be directed, at least in part, to an external or shared data source or data repository.

[0179] In some embodiments, a representation of the diffraction information may be output as part of a voxSceneDiffractionPreComputedPathData() syntax element in accordance with ISO / IEC 23090-4 (Coded representation of immersive media—Part 4: MPEG-I Immersive Audio, https: / / www.iso.org / standard / 84711.html) or any future standard derived therefrom.

[0180] For example, Table 2 may give the syntax elements of voxSceneDiffractionMap().

[0181] Table 2—Syntax for voxSceneDiffractionMap()

[0182]

[0183] voxSceneDiffractionMap() provides a compact representation of a 2D diffraction map (VoxDataDiffractionMap). This 2D representation is similar to the 3D representation used for voxel-based 3D audio scenes.

[0184] A MapElement is defined by two points (x,y index) on the diffraction map and the corresponding value. The two points span the rectangle, and all covered grid cells are assigned the value voxDiffractionMapValue.

[0185] The bitstream element numberOfVoxDiffractionMapElements indicates the number of MapElements.

[0186] The bitstream element voxDiffractionMapValue_ represents a binary value that controls the pathfinding algorithm. This is useful because the value indicates whether a path can pass through a grid cell. All entries on the diffraction map have this value defined.

[0187] The bitstream element voxDiffractionMapPosPackedS represents a packed representation of 2 indices of the starting grid cell of a MapElement. It can be an array showing the set of all starting grid cells.

[0188] The bitstream element voxDiffractionMapPosPackedE represents a packed representation of 2 indexes of the end grid cell of a MapElement. It can be an array showing the set of all end grid cells.

[0189] Both voxDiffractionMapPosPackedS and voxDiffractionMapPosPackedE are useful because they allow for a compact representation of the data, where a single voxDiffractionMapValue is used for all grid cells between these two variables.

[0190] Figure 4 is a flow chart illustrating an example of a method 400, which includes steps that may be performed to implement the steps of method 300. Method 400 includes steps S410 to S460. Among them, steps S410 to S450 may implement step 330 of method 300. Further, step S460 may correspond to step S350.

[0191] exist Step S410 At the audio scene description and the listener position, the current scene state is determined.

[0192] exist Step S420 At, determining whether the current scene state corresponds to a known scene state for which pre-computed diffraction information is available (e.g., can be retrieved). For example, the pre-computed diffraction information can be retrieved from a bitstream (incoming bitstream) or a storage device (particularly including an external or shared storage device). Determining whether the current scene state corresponds to a known scene state may include determining a hash value based on the current scene state. It may further include comparing the hash value of the current scene state with a hash value of a known (e.g., previously encountered) scene state.

[0193] If it is determined that the current scene state corresponds to a known scene state (in Step S430 If the answer is yes), the method proceeds to step S440.

[0194] exist Step S440 At the scene state, the diffraction information is determined by extracting pre-computed diffraction information of a known scene state from a bitstream or a storage device. The storage device may involve a local storage device (e.g., a memory, a cache, a file, etc.) or a shared storage device (e.g., a cloud storage device, a server storage device).

[0195] Extracting pre-computed diffraction information for known scene states may include receiving a lookup table or entries of the lookup table from a bitstream (incoming bitstream) or a storage device. The lookup table may be viewed as a representation of the diffraction information. It may include multiple items of pre-computed diffraction information, each item of pre-computed diffraction information being associated with a corresponding known scene state. The pre-computed diffraction information and the associated known scene state may correspond to the above-mentioned data elements. The known scene state may include or indicate a known audio scene description and a known listener position.

[0196] As described above, selecting the relevant entry of a received lookup table, or selecting the relevant entry to be received (if not the entire lookup table but only its entries) may involve using a hash value.

[0197] On the other hand, if it is determined that the current scene state does not correspond to a known scene state (in Step S430 If the answer is No), the method proceeds to step S450.

[0198] exist Step S450 At , a pathfinding algorithm is used based on the source position, the listener position and the representation of the three-dimensional audio scene to determine the diffraction information. This can be done according to the above reference Figure 2 The procedure described is completed.

[0199] exist Step S460 At step S440 or step S450, the diffraction information obtained is output. This step may correspond to the above step S350.

[0200] In summary, the proposed method may include (among other things) the following:

[0201] • Check whether pre-computed diffraction path information (ie, pre-computed diffraction information) can be fetched from cache and re-applied to the current scene state and listener position.

[0202] • Check whether pre-computed diffraction path information (ie, pre-computed diffraction information) for the current scene state is available from a memory cache or a bitstream and re-applied.

[0203] The scene state is obtained by the input parameter L of the function DiffractionDirectionCalculation(). vox , S vox , VoxDataDiffractionMap, which includes the pathfinding algorithm, voxel C vox Selection and diffraction path length estimation steps.

[0204] The diffraction path information (eg, diffraction information) is obtained via the output parameter C vox 、r in The diffraction path information (if available) can be obtained directly from the bitstream syntax voxSceneDiffractionPreComputedPathData() of the corresponding scene state to avoid the function DiffractionDirectionCalculation() call.

[0205] When the current scene state L vox , S vox, VoxDataDiffractionMap obtains diffraction path information C vox 、r in , this information can be cached in memory (and provided outside the renderer) for later reuse by the renderer (or other renderer instances).

[0206] In other words, “diffraction pathfinding” according to the present disclosure (e.g., as embodied by method 300 and / or method 400) may involve the following processing:

[0207] - Check whether pre-computed diffraction path information for the current scene state can be obtained from the memory cache or bitstream and re-applied.

[0208] The scene state is calculated via the input parameter L of the function DiffractionDirectionCalculation(). vox , S vox , VoxDataDiffractionMap, which includes the pathfinding algorithm, voxel C vox Selection and diffraction path length estimation steps.

[0209] [C vox ,r in ]=DiffractionDirectionCalculation(L vox ,S vox ,VoxDataDiffractionMap)

[0210] The diffraction path information is obtained via the output parameter C vox 、r in The diffraction path information (if available) can be obtained directly from the bitstream syntax voxSceneDiffractionPreComputedPathData() of the corresponding scene state to avoid the function DiffractionDirectionCalculation() call.

[0211] -When the current scene state L vox , S vox , VoxDataDiffractionMap obtains diffraction path information C vox 、r in , this information can be cached in memory (and provided outside the renderer) for later reuse by the renderer (or other renderer instances).

[0212] In the above, the bitstream syntax definition can be written in the style of a function() in the MPEG standard document. It defines how to read / parse data (bitstream elements) from the bitstream. In this case, it is used to obtain the necessary variables / information to recover the diffraction path information.

[0213] Figure 5 , Figure 6 and Figure 7 An example of a processing chain 500 according to the above is shown, which can be used to process audio scene information for audio rendering. Specifically, the processing chain 500 can be used to convert voxel related data into parameters and signals required for auralization.

[0214] Figure 5 It involves the case where the current scene state is an unknown scene state. As mentioned above, different from Figure 1 In the processing chain 100 of FIG. 1 , the bitstream / memory 510 additionally comprises data elements, for example in the form of a look-up table, containing diffraction information and associated scene states.

[0215] Like the processing chain 100, the processing chain 500 receives an audio scene description 20 from a bitstream (or storage / memory) 510. The processing chain 500 further receives an indication of a user position (listener position) 30 of a user (listener) within the audio scene.

[0216] The diffraction direction calculation block (diffraction calculation block) 40 for determining (eg calculating) diffraction information and the diffraction modeling tool 50 for applying diffraction modeling and optionally occlusion modeling based on the diffraction information may be identical to that of the processing chain 100 .

[0217] However, unlike Figure 15. The processing chain 100 of FIG. 10 uses the audio scene description 20 and the listener position 30 to determine a scene state 515 or a scene state identifier. This scene state 515 (e.g., scene state N1 defined above) or scene state identifier (e.g., HASH(N1)) is provided / input to a scene state analysis block 520, which determines whether the current scene state 515 corresponds to a known scene state 530 (yes at block 535) or not (no at block 535). In this example, the current scene state 515 does not correspond to a known scene state (i.e., the current scene state 515 is an unknown scene state). Therefore, the audio scene description 20 and the listener position 30 are input to the diffraction direction calculation block 40 in the same manner as the processing chain 100 to generate diffraction information 550. Subsequently, the diffraction information 550 is used for rendering / diffraction modeling, just as in the case of the processing chain 100. However, in addition, the diffraction information 550 (e.g., the diffraction information N2 or its quantized version N3 as defined above) is output via an interface for later reuse by a renderer or other (external) rendering instance. Specifically, the diffraction information 550 can be output to a bitstream (or memory / storage device) 510 via an output interface 555 together with the corresponding scene state.

[0218] Figure 6 This involves the case where the current scene state 515 is a known scene state.

[0219] Likewise, the current scene state 515 is provided / input to the scene state analysis block 520 to determine whether the current scene state 515 corresponds to a known scene state 530. In the present example, the current scene state 515 corresponds to a known scene state. Thus, instead of inputting the audio scene description 20 and the listener position 30 to the diffraction direction calculation block 40 to calculate / generate the diffraction information, the diffraction information is extracted / received from the bitstream (or storage device / memory) 510 as described above (e.g., via step S450 of method 400). However, even if the diffraction information is not calculated locally, it may be output to the bitstream (or memory / storage device) 510, just as Figure 5 The reason is that if diffraction information is available from one source (e.g. in a bitstream or a storage device), it can be made available to the corresponding other source(s) in this way.

[0220] Figure 7 The complete processing chain 500 is shown, including the data paths for known and unknown scene states 515 .

[0221] Figure 8810 is a diagram illustrating the complexity measure of different embodiments of processing audio scene information or audio rendering when assuming a simple maze is an audio scene. It is further assumed that the user randomly traverses the maze, thereby revisiting previously visited locations. Graph 810 relates to a situation where no pre-calculated diffraction information is available (e.g., no diffraction information is provided, and the bitstream, memory / cache are disabled). In this case, the computational load on the renderer is substantially constant and relatively high. Graph 820 relates to a situation where pre-calculated diffraction information is available locally (e.g., no diffraction information is provided, and the bitstream, local memory / cache are enabled). In this case, the processing load on the renderer decays over time because more and more diffraction information items are accumulated locally. In other words, more and more encountered scene states will be related to (local) known scene states. Finally, graph 830 relates to a situation where pre-calculated diffraction information is provided externally (e.g., complete diffraction information is provided via a bitstream). In this case, the computational load on the renderer remains low because important parts of the scene state are related to known scene states and the diffraction information can be obtained externally (e.g., from a bitstream or by request from external / shared storage) without having to be calculated locally.

[0222] Fig. 9 An example of a possible use case of the technology according to an embodiment of the present disclosure is schematically illustrated. Two listeners (users) A and B at different locations within an audio scene (e.g., a house with different areas and floors) are shown. Users A and B may be users who explore a VR environment including an audio scene individually or together, for example as part of a game, a virtual tour, etc. For example, the users exploring a common VR environment may be running social VR. Users A and B have different listener positions within the audio scene, and therefore will produce different rendering results and different diffraction information. The present disclosure foresees that each user (or its corresponding device / decoder / renderer) makes its calculated diffraction information available to other users. Once user B enters an area in the audio scene where user A has previously appeared, they can benefit from the diffraction information pre-calculated by user A, and vice versa. For example, the diffraction information of user A can be provided to user B via a LUT, which indexes different diffraction information items with corresponding scene states or scene state identifiers. By exchanging diffraction information between different devices / decoders / renderers, the computational load of the devices / decoders / renderers of the two users can be reduced according to their movement patterns within the audio scene.

[0223] Additionally, since users (listeners) tend to behave similarly, diffraction information (diffraction data) accumulates, especially for relevant (e.g., frequently occurring) scene states. This is very difficult to implement for encoder-side pre-computation of the diffraction information, since the encoder has no access to the actual listener positions and can therefore only make assumptions about them. Further, since part of the pre-computed diffraction information in this case is related to irrelevant or less relevant scene states, the use of data storage (e.g., physical / shared storage or bitstream bandwidth) is much less efficient for encoder-side pre-computation.

[0224] For example, the proposed functions and techniques may create a LUT corresponding to the user's real 6DoF behavior (rather than assumed behavior on the encoder side), and may therefore be said to involve intelligent user-oriented LUT creation.

[0225] Efficient voxel-based representation of audio scenes

[0226] For example, current representation formats for voxel-based scenes include *.vox, *.binvox, and the like.

[0227] The present disclosure provides the following compression method for the MPEG-I audio standard, for example. A group of voxels having the same acoustic properties (e.g., the same material properties) or the same audio rendering instructions can be identified by forming two points of a cubic region on a voxel grid. All voxels in the cubic region share the acoustic properties or audio rendering instruction sets assigned to the corresponding two points, as shown below.

[0228] That is, the scene geometry is determined by a set of point pairs like this (as an example of a geometric region representation), for example:

[0229] <VoxBox id="V_ID"material="P_ID"Point_S="X1 Y1 Z1"Point_E="X2 Y2Z2" / >

[0230] Where V_ID is the cubic voxel block element identifier; P_ID is the acoustic property or audio rendering instruction set identifier (e.g., occlusion, reflection, RT60 data); X1, Y1, Z1 and X2, Y2, Z2 are the grid indices that define the corresponding two points of the cubic voxel block element. Accordingly,<VoxBox id,material,Point_S,Point_E / > may correspond to the scene elements defined above, ie a geometric region (eg defined by point pairs) together with its voxel attributes (eg P_ID) and optionally its identifier (eg V_ID).

[0231] The voxel size can be determined by the number of voxels, as:

[0232] <VoxSize size="N" / >

[0233] where N is the number of voxels along the first longest scene dimension.

[0234] Accordingly, a voxel-based representation of an audio scene according to an embodiment of the present disclosure may include representations or indications or equivalent rendering instructions of one or more cubic geometric regions (cubic spatial regions, cubic volumes) having equivalent (i.e., identical, common) acoustic properties (e.g., materials, absorption coefficients, reflection coefficients, etc.). The acoustic properties or rendering instructions of a given voxel may be non-limiting examples of voxel properties of a given voxel. For example, a representation of an indication of a cubic geometric region may involve scene elements. It should be understood that the cubic regions are each important because they each include more than a single voxel and are composed of a set of connected (i.e., continuous) voxels.

[0235] The shape of each of these geometric regions can be defined by a first boundary voxel and a second boundary voxel (e.g., the point pair above). For example, these first boundary voxels and second boundary voxels can be related to the diameter angles (polar angle voxels) of the cube, such as the polar angle voxel with the smallest x, y, z coordinate value or coordinate index and the polar angle voxel with the largest x, y, z coordinate value or coordinate index. Other choices of diameter polar angle voxels are also feasible.

[0236] Further, in the voxel-based representation, each geometric region may be represented by at least an indication of first and second boundary voxels and an indication of common voxel properties of the voxels within the geometric region. Additionally, the representation of the geometric region may include an identifier (ID) of the geometric region.

[0237] Fig. 10A An example of a geometric region 1002 in a voxel grid 1007 is shown. The geometric region 1002 includes a plurality of voxels 1003 forming a cube. The shape or size of the geometric region can be represented by a first boundary voxel 1004-1 and a second boundary voxel 1004-2 (such as polar corners or polar angle voxels), which in this example correspond to the front lower left corner and the back upper right corner, respectively. Fig. 10B Shows from Fig. 10A A side view of geometric area 1002 viewed from the right front.

[0238] For the proposed compression method, each next pair of points (i.e., each next geometric region) can redefine the voxel attributes in the corresponding cubic region. That is, the voxel attributes of the subsequent geometric region can override any voxel attributes previously assigned to the voxels of the subsequent geometric region. In one embodiment, a smaller geometric region that is completely contained within a larger geometric region can redefine or override the voxel attributes of the voxels in the smaller geometric region with the voxel attributes of the smaller geometric region. Here, it should be understood that the corresponding voxel attributes are overwritten while any other voxel attributes are retained. For example, if a subsequent geometric region defines acoustic properties for its voxels, these acoustic properties will be used to override the acoustic properties defined for the voxels of the previous geometric region, but any rendering instructions for the voxels of the previous geometric region will be retained.

[0239] As described above, for example, the order among geometric regions can be derived from whether the geometric regions are completely contained within each other, can be derived from the order of representation or indication of the geometric regions in the bitstream, or can be derived from a predefined order of identifiers (IDs) of reference geometric regions.

[0240] Fig. 10C It is schematically illustrated how voxel properties assigned to a first geometric region defined by polar voxels 1004 - 1 , 1004 - 2 may be overwritten by voxel properties of a second geometric region 1002 .

[0241] Fig.11 An example of a voxel-based description of a three-dimensional audio scene is schematically illustrated, the three-dimensional audio scene comprising a plurality of cubic volumes having common voxel properties, the plurality of cubic volumes potentially having smaller cubic sub-volumes that redefine the voxel properties. The audio scene may include voxels 1101, which are local sound occluders that occlude within a given acoustic environment of the audio scene. The audio scene may additionally include voxels 1102, which are local sound occluders that separate acoustic environments from each other.

[0242] Representation of voxel coordinates / indexes

[0243] For example, the following efficient representation of voxel indices can be suitable for transmission or storage of voxel grids and diffraction map (VoxDataDiffractionMap) entries. It can replace any fixed-length representation of voxel indices (voxel coordinates).

[0244] The following steps can be performed in the context of the proposed representation:

[0245] Step 1: Determine the number of bits (i.e., quantity, count) required for the current grid resolution / diffraction map dimension. For a 3D voxel grid and a 2D diffraction map, for example, these quantities NbitsVox and NbitsMap, respectively, can be determined as follows:

[0246] NbitsVox=ceil(log2(L*W*H-1))

[0247] NbitsMap=ceil(log2(L*W-1)),

[0248] Where L, W, H (length, width, height) are the dimensions of the voxel grid and the diffraction pattern. The values ​​of the voxel grid and the diffraction pattern may be different.

[0249] Step 1 can be applicable to both the encoder side and the decoder side.

[0250] Step 2: Map the voxel index (x, y, z) and the diffraction pattern index (x, y) to the packed representation index (Idx) and encode them using Nbits_vox and Nbits_map bits, respectively. In one embodiment, (x, y, z) is zero-based and the packed representation index may range from 0 to L*W*H-1 for voxels and from 0 to L*W-1 for diffraction patterns. For example, the mapping from the index (x, y, z) to the packed representation index may be as follows:

[0251] Idx(x,y,z)=(x-1)+((y-1)*L)+((z-1)*L*W).

[0252] Step 2 can be performed only on the encoder side.

[0253] In the above, a packed representation index is an index that can uniquely identify a voxel in a voxel grid or a diffraction map. In other words, a voxel in a voxel grid can be assigned a unique continuous index so that each voxel in the voxel grid can be uniquely identified by a single integer. Accordingly, a packed representation index can be used for any indication of a voxel position in a voxel grid or a two-dimensional map. In particular, a packed representation index can be used to indicate any voxel position mentioned throughout the disclosure.

[0254] Unique indices may be assigned to voxels according to a predefined pattern. For example, the voxel grid may be scanned / traversed in this order in the x, y, and z directions to successively assign unique indices to corresponding voxels.

[0255] For example, the mapping from packed representation indices to voxel and diffraction pattern indices can be as follows:

[0256] x = floor (Idx) % L + 1

[0257] y=floor(Idx / L)%W+1

[0258] z=floor(Idx / L / W)%H+1,

[0259] Here, % represents the modulo operator.

[0260] Based on the above, the bitstream payload element voxSceneDiffractionPreComputedPathData() given in Table 1 can use the following pseudo code:

[0261]

[0262]

[0263] Among them, voxDiffractionMapPosPackedS and voxDiffractionMapPosPackedE indicate the packed representation index.

[0264] Entropy Coding

[0265] Entropy encoding methods may be applied to sequences of integers representing acoustic properties, voxel grid coordinates, voxel grid indices, and the like.

[0266] For example, an entropy coding method may be applied to the above integer sequence (representing acoustic properties or audio rendering instruction set reference P_ID and grid indexes X1, Y1, Z1 and X2, Y2, Z2) or its packed representation. Further, entropy coding may be applied to the integer sequence obtained from the above representation of the diffraction path information.

[0267] Accordingly, the technical benefits and advantages of voxel-based scene representation as described herein are to allow complex dynamic scenes to be created and efficiently encoded.

[0268] Device

[0269] Although methods and processing chains have been described above, it should be understood that the present disclosure also relates to devices (eg computer devices or devices having processing capabilities in general) for implementing these methods and processing chains (or general techniques).

[0270] exist Fig.12 An example of such an apparatus 1200 is schematically illustrated in . The apparatus 1200 includes a processor 1201 and a memory 1202 coupled to the processor 1201. The memory 1202 may store instructions for execution by the processor 1201. The processor 1201 may be suitable for implementing the processing chain described throughout the disclosure and / or performing the method described throughout the disclosure (e.g., a method of processing audio scene information for audio rendering). The apparatus 1200 may receive input (e.g., an audio scene description, a listener position, etc.) and generate output (e.g., a representation of diffraction information, etc.).

[0271] explain

[0272] Aspects of the systems described herein can be implemented in an appropriate computer-based sound processing network environment (e.g., a server or cloud environment) to process digital or digitized audio files. Portions of these systems may include one or more networks comprising any desired number of independent machines including one or more routers (not shown) for buffering and routing data transmitted between computers. Such networks can be built on a variety of different network protocols and can be the Internet, a wide area network (WAN), a local area network (LAN), or any combination thereof.

[0273] One or more components, blocks, processes or other functional components may be implemented by a computer program executed by a processor-based computing device that controls the system. It should also be noted that any number of combinations of hardware, firmware and / or data and / or instructions embodied in various machine-readable or computer-readable media may be used to describe the various functions disclosed herein from the perspective of their behavior, register transfers, logic components and / or other features. Computer-readable media that may embody such formatted data and / or instructions include, but are not limited to, various forms of physical (non-transitory), non-volatile storage media, such as optical, magnetic or semiconductor storage media.

[0274] In particular, it should be understood that embodiments may include hardware, software, and electronic components or modules, and for discussion purposes, may be illustrated and described as if most components were implemented only in hardware. However, one of ordinary skill in the art, and based on a reading of this detailed description, will recognize that in at least one embodiment, the electronic-based aspects may be implemented in software (e.g., stored on a non-transitory computer-readable medium) that is executable by one or more electronic processors (such as microprocessors and / or application specific integrated circuits ("ASICs")). Therefore, it should be noted that embodiments may be implemented using a variety of hardware and software-based devices and a variety of different structural components. For example, the computer-implemented neural network described herein may include one or more electronic processors, one or more computer-readable media modules, one or more input / output interfaces, and various connections (e.g., a system bus) connecting the various components.

[0275] Although one or more embodiments have been described by way of example and with respect to specific embodiments, it should be understood that one or more embodiments are not limited to the disclosed embodiments. On the contrary, it is intended to cover various modifications and similar arrangements that are obvious to those skilled in the art. Therefore, the scope of the appended claims should be given the broadest interpretation so as to cover all such modifications and similar arrangements.

[0276] Likewise, it should be understood that the phraseology and terminology used herein is for descriptive purposes and should not be regarded as limiting. The use of "including," "comprising," or "having" and variations thereof are meant to encompass the items listed thereafter and equivalents thereof as well as additional items. Unless otherwise specified or limited, the terms "mounted," "connected," "supported," and "coupled" and variations thereof are used broadly and encompass direct and indirect mountings, connections, supports, and couplings.

[0277] Enumerated example embodiments

[0278] Various aspects and embodiments of the present invention may also be understood from the following enumerated example embodiments (EEE), which are not claims.

[0279] EEE1. A method for processing audio scene information for audio rendering, the method comprising:

[0280] receiving an audio scene description comprising a representation of a three-dimensional audio scene and information about source positions of sound sources within the audio scene;

[0281] receiving an indication of a listener position of a listener within the audio scene;

[0282] obtaining diffraction information associated with an acoustic diffraction path between the source position and the listener position within the audio scene;

[0283] performing audio rendering on the sound source based on the diffraction information; and

[0284] A representation of the diffraction information is output.

[0285] EEE2. A method according to EEE1, wherein outputting a representation of the diffraction information comprises outputting a data element comprising the diffraction information and information about a scene state, the scene state comprising the audio scene description and the listener position. EEE3.

[0286] EEE3. A method according to EEE1 or EEE2, wherein the representation of the diffraction information is output to a bit stream and / or a storage device.

[0287] EEE4. A method according to any one of EEE1 to EEE3, wherein the diffraction information is output for later reuse in audio rendering by the same rendering instance, or for later reuse by another rendering instance. EEE4.

[0288] EEE5. A method according to any one of EEE1 to EEE4, wherein the representation of the diffraction information is output as part of a voxSceneDiffractionPreComputedPathData() syntax element according to ISO / IEC 23090-4 or any standard derived from ISO / IEC 23090-4.

[0289] EEE6. A method according to any one of EEE1 to EEE5, wherein the diffraction information indicates a virtual source position of a virtual sound source.

[0290] EEE7. A method according to any one of EEE1 to EEE6, wherein the representation of the three-dimensional audio scene is a voxel-based representation; and

[0291] wherein the representation of the three-dimensional audio scene comprises one or more indications of cubic volumes in a voxel grid, and wherein each such indication comprises information about a pair of polar voxels of the cubic volume and information about common voxel properties of the voxels in the cubic volume.

[0292] EEE8. A method according to EEE7, wherein the information about the pair of polar angle voxels of the cubic volume includes an indication of the corresponding voxel index assigned to the polar angle voxel, and the voxels of the voxel-based audio scene representation have uniquely assigned continuous voxel indices.

[0293] EEE9. A method according to any one of EEE1 to EEE6, wherein the representation of the three-dimensional audio scene is a voxel-based representation;

[0294] wherein the diffraction information includes an indication of the location of voxels located on or near the diffraction path and an indication of the length of the diffraction path; and

[0295] Wherein, the indication of the position of a voxel located on or near the diffraction path is an indication of a voxel index assigned to the voxel, and voxels of the voxel-based audio scene representation have uniquely assigned consecutive voxel indices.

[0296] EEE10. The method according to any one of EEE1 to EEE9, further comprising:

[0297] A current scene state is determined based on the audio scene description and the listener position.

[0298] EEE11. The method according to EEE10 further comprises:

[0299] A determination is made as to whether the current scene state corresponds to a known scene state for which pre-computed diffraction information is available.

[0300] EEE12. A method according to EEE11, wherein determining whether the current scene state corresponds to a known scene state comprises determining a hash value based on the current scene state.

[0301] EEE13. The method according to EEE11 or EEE12 further comprises:

[0302] If it is determined that the current scene state corresponds to a known scene state, the diffraction information is determined by extracting pre-computed diffraction information for the known scene state from a bitstream or a storage device.

[0303] EEE14. The method according to any one of EEE11 to EEE13, further comprising:

[0304] If it is determined that the current scene state does not correspond to a known scene state, the diffraction information is determined using a pathfinding algorithm based on the source positions, the listener positions and the representation of the three-dimensional audio scene.

[0305] EEE15. The method according to any one of EEE1 to EEE14 further includes: receiving a lookup table or an entry of a lookup table from a bitstream or a storage device, wherein the lookup table includes multiple pre-calculated diffraction information, each pre-calculated diffraction information is associated with a corresponding known scene state, and the known scene state includes a known audio scene description and a known listener position.

[0306] EEE16. A method according to any one of EEE1 to EEE15, wherein the representation of the three-dimensional audio scene is a voxel-based representation.

[0307] EEE17. A method for compressing an audio scene for three-dimensional audio rendering, the method comprising:

[0308] obtaining a voxelized representation of the audio scene, the voxelized representation comprising a plurality of voxels arranged in a voxel grid, each voxel having associated voxel attributes;

[0309] determining, among the voxels of the voxelized representation, a set of voxels that form a connected geometric region on the voxel grid, wherein the voxels in the geometric region share common voxel properties; and

[0310] A representation of the audio scene is generated based on the determined set of voxels.

[0311] EEE18. A method according to EEE17, wherein the geometric region has a cubic shape, and the method further includes determining at least a first boundary voxel and a second boundary voxel of the group of voxels from the multiple voxels represented by the voxelization, wherein the first boundary voxel and the second boundary voxel define the cubic shape of the geometric region.

[0312] EEE19. A method according to EEE17 or EEE18, wherein the voxel properties of each voxel include acoustic properties associated with the voxel and / or a set of audio rendering instructions assigned to the voxel, and the common voxel properties of the voxels in the geometric region include common acoustic properties associated with those voxels and / or a set of common audio rendering instructions assigned to those voxels.

[0313] EEE20. The method according to any one of EEE17 to EEE19 further includes determining at least one scene element parameter for the geometric area, the at least one scene element parameter comprising one or more of the following: a scene element identifier, an acoustic property identifier and / or an audio rendering instruction set identifier, and an index of a corresponding first boundary voxel and a second boundary voxel that define the geometric area.

[0314] EEE21. The method according to EEE20 further includes applying entropy coding to at least one scene element parameter of the geometric area.

[0315] EEE22. The method according to EEE20 or EEE21 further includes outputting a bitstream including the at least one scene element parameter to determine the set of voxels associated with the geometric region of the compressed representation of the audio scene based on the determined set of voxels.

[0316] EEE23. A method according to any one of EEE17 to EEE22, wherein the geometric area is related to a scene element within the audio scene.

[0317] EEE24. A method according to any one of EEE17 to EEE23, wherein the audio scene includes a large scene represented by a determined set of voxels, the large scene includes a set of sub-scenes, wherein each of the sub-scenes corresponds to a subset of the determined set of voxels, and the method further includes determining a subset of voxels corresponding to the sub-scene in the determined set of voxels.

[0318] EEE25. A method according to any one of EEE17 to EEE24, further comprising applying interpolation to audio voxels in time and / or space.

[0319] EEE26. The method according to any one of EEE17 to EEE25 further includes redefining voxel attributes of a subset of the set of voxels associated with the scene sub-elements in the geometric region for covering the subset with the redefined voxel attributes.

[0320] EEE27. According to any one of the methods described in EEE17 to EEE26, the method further includes determining a voxel superset including the determined set of voxels, the determined set of voxels being associated with scene sub-elements within the geometric region, and the method further includes assigning new voxel attributes to the determined set of voxels and overwriting the voxel attributes of the determined set of voxels with the new voxel attributes.

[0321] EEE28. The method according to any one of EEE17 to EEE27 further includes determining a voxel size used to represent the geometric region, wherein the voxel size is based on the number of voxels along a scene dimension of the geometric region.

[0322] EEE29. An apparatus comprising a processor and a memory, the memory being coupled to the processor and storing instructions for the processor, wherein the processor is adapted to execute a method according to any one of EEE1 to EEE28.

[0323] EEE30. A program comprising instructions which, when executed by a processor, cause the processor to perform a method according to any one of EEE1 to EEE28.

[0324] EEE31. A computer-readable storage medium storing the program according to EEE30.

Claims

1. A method (300) for processing audio scene information for audio rendering, the method (300) comprising: receiving (S310) an audio scene description, the audio scene description comprising a representation of a three-dimensional audio scene and information about source positions (220) of sound sources within the audio scene; receiving (S320) an indication of a listener position (210) of a listener within the audio scene; Obtaining (S330) diffraction information associated with an acoustic diffraction path between the source position and the listener position within the audio scene; Performing (S340) audio rendering on the sound source based on the diffraction information; as well as A representation of the diffraction information is output (S350).

2. The method (300) according to claim 1, wherein: Outputting (S350) the representation of the diffraction information comprises outputting a data element comprising the diffraction information and information about a scene state, the scene state comprising the audio scene description and the listener position (210).

3. The method (300) according to any one of the preceding claims, wherein: The representation of the diffraction information is output to a bitstream and / or to a storage device.

4. A method according to any one of the preceding claims, wherein: The diffraction information is output for later reuse in audio rendering by the same rendering instance, or for later reuse by another rendering instance.

5. The method (300) according to any one of the preceding claims, wherein: According to standards set by ISO / IEC, such as the MPEG-I standard, the representation of the diffraction information is output as part of the voxSceneDiffractionPreComputedPathData() syntax element.

6. The method (300) according to any one of the preceding claims, wherein: The diffraction information indicates a virtual source position of a virtual sound source.

7. The method (300) according to any one of the preceding claims, wherein: The representation of the three-dimensional audio scene is a voxel-based representation; and wherein the representation of the three-dimensional audio scene comprises one or more indications of cubic volumes in a voxel grid, and wherein each such indication comprises information about a pair of polar voxels of the cubic volume and information about common voxel properties of the voxels in the cubic volume.

8. The method (300) of claim 7, wherein: The information about the pair of polar angle voxels of the cubic volume comprises an indication of respective voxel indices assigned to the polar angle voxels, the voxels of the voxel-based audio scene representation having uniquely assigned consecutive voxel indices.

9. The method (300) according to any one of claims 1 to 6, wherein: The representation of the three-dimensional audio scene is a voxel-based representation; wherein the diffraction information includes an indication of the location of voxels located on or near the diffraction path and an indication of the length of the diffraction path; and Wherein, the indication of the position of a voxel located on or near the diffraction path is an indication of a voxel index assigned to the voxel, and voxels of the voxel-based audio scene representation have uniquely assigned consecutive voxel indices.

10. The method (300) according to any one of the preceding claims, further comprising: A current scene state is determined based on the audio scene description and the listener position.

11. The method (300) of claim 10, further comprising: A determination is made as to whether the current scene state corresponds to a known scene state for which pre-computed diffraction information is available.

12. The method (300) of claim 11, wherein: Determining whether the current scene state corresponds to a known scene state includes determining a hash value based on the current scene state.

13. The method (300) according to claim 11 or 12, further comprising: If it is determined that the current scene state corresponds to a known scene state, the diffraction information is determined by extracting pre-computed diffraction information for the known scene state from a bitstream or a storage device.

14. The method (300) according to any one of claims 11 to 13, further comprising: If it is determined that the current scene state does not correspond to a known scene state, the diffraction information is determined using a pathfinding algorithm based on the source positions, the listener positions and the representation of the three-dimensional audio scene.

15. The method (300) according to any one of the preceding claims, further comprising: A lookup table or an entry of the lookup table is received from a bitstream or a storage device, the lookup table comprising a plurality of pre-computed diffraction information, each pre-computed diffraction information being associated with a corresponding known scene state, the known scene state comprising a known audio scene description and a known listener position.

16. The method (300) according to any one of the preceding claims, wherein: The representation of the three-dimensional audio scene is a voxel-based representation including information about positions and material properties of a plurality of occlusion voxels.

17. A method for compressing an audio scene for three-dimensional audio rendering, the method comprising: obtaining a voxelized representation of the audio scene, the voxelized representation comprising a plurality of voxels arranged in a voxel grid, each voxel having associated voxel attributes; determining, among the voxels of the voxelized representation, a set of voxels that form a connected geometric region on the voxel grid, wherein the voxels in the geometric region share common voxel properties; as well as A representation of the audio scene is generated based on the determined set of voxels.

18. The method according to claim 17, wherein: The geometric region has a cubic shape, and the method further includes determining at least a first boundary voxel and a second boundary voxel of the set of voxels from the plurality of voxels of the voxelized representation, the first boundary voxel and the second boundary voxel defining the cubic shape of the geometric region.

19. The method according to claim 17 or 18, wherein: The voxel properties of each voxel include acoustic properties associated with the voxel and / or a set of audio rendering instructions assigned to the voxel, and the common voxel properties of the voxels in the geometric region include common acoustic properties associated with those voxels and / or a set of common audio rendering instructions assigned to those voxels.

20. The method according to any one of claims 17 to 19 further comprises determining at least one scene element parameter for the geometric region, the at least one scene element parameter comprising one or more of the following: a scene element identifier, an acoustic property identifier and / or an audio rendering instruction set identifier, and an index of a corresponding first boundary voxel and a second boundary voxel defining the geometric region.

21. The method of claim 20, further comprising applying entropy coding to the at least one scene element parameter of the geometric region.

22. The method of claim 20 or 21, further comprising outputting a bitstream comprising the at least one scene element parameter to determine a set of voxels associated with a geometric region of the compressed representation of the audio scene based on the determined set of voxels.

23. The method according to any one of claims 17 to 22, wherein: The geometric regions are associated with scene elements within the audio scene.

24. The method according to any one of claims 17 to 23, wherein: The audio scene includes a large scene represented by a determined set of voxels, the large scene includes a set of sub-scenes, wherein each of the sub-scenes corresponds to a subset of the determined set of voxels, and the method further includes determining a subset of voxels corresponding to the sub-scene in the determined set of voxels.

25. A method according to any one of claims 17 to 24, further comprising applying interpolation to the audio voxels in time and / or space.

26. The method according to any one of claims 17 to 25, further comprising redefining voxel properties of a subset of the set of voxels associated with the scene sub-element in the geometric region for overwriting the subset with the redefined voxel properties.

27. The method according to any one of claims 17 to 26 further comprises determining a voxel superset including the determined set of voxels, the determined set of voxels being associated with scene sub-elements within the geometric region, the method further comprising assigning new voxel attributes to the determined set of voxels and overwriting the voxel attributes of the determined set of voxels with the new voxel attributes.

28. The method of any one of claims 17 to 27, further comprising determining a voxel size for representing the geometric region, wherein: The voxel size is based on the number of voxels along a scene dimension of the geometric region.

29. An apparatus (1000), comprising a processor (1010) and a memory (1020), the memory being coupled to the processor and storing instructions for the processor (1010), wherein: The processor (1010) is adapted to perform the method according to any one of claims 1 to 28.

30. A program comprising instructions which, when executed by a processor, cause the processor to perform the method according to any one of claims 1 to 28.

31. A computer-readable storage medium storing the program according to claim 30.