METHOD AND SYSTEM FOR CODING METADATA IN AUDIO DATA STREAMS AND FOR FLEXIBLE INTRAOBJECT AND INTEROBJECT BITRATE ADJUSTMENT

Description

TECHNICAL FIELD

[0001] The present disclosure relates to sound coding, more specifically to a technique for digitally coding object-based audio, for example speech, music or general audio sound. In particular, the present disclosure relates to a system and method for coding and a system and method for decoding an object-based audio signal comprising audio objects in response to audio streams with associated metadata.

[0002] In the present disclosure and the appended claims: (a) The term "object-based audio" is intended to represent a complex audio auditory scene as a collection of individual elements, also known as audio objects. Also, as indicated herein above, "object-based audio" may comprise, for example, speech, music or general audio sound. (b) The term "audio object" is intended to designate an audio stream with associated metadata. For example, in the present disclosure, an "audio object" is referred to as an independent audio stream with metadata (ISm). (c) The term "audio stream" is intended to represent, in a bit-stream, an audio waveform, for example speech, music or general audio sound, and may consist of one channel (mono) though two channels (stereo) might be also considered. "Mono" is the abbreviation of "monophonic" and "stereo" the abbreviation of "stereophonic." (d) The term "metadata" is intended to represent a set of information describing an audio stream and an artistic intension used to translate the original or coded audio objects to a reproduction system. The metadata usually describes spatial properties of each individual audio object, such as position, orientation, volume, width, etc. In the context of the present disclosure, two sets of metadata are considered: input metadata: unquantized metadata representation used as an input to a codec; the present disclosure is not restricted a specific format of input metadata; and coded metadata: quantized and coded metadata forming part of a bit-stream transmitted from an encoder to a decoder. (e) The term "audio format" is intended to designate an approach to achieve an immersive audio experience. (f) The term "reproduction system" is intended to designate an element, in a decoder, capable of rendering audio objects, for example but not exclusively in a 3D (Three-Dimensional) audio space around a listener using the transmitted metadata and artistic intension at the reproduction side. The rendering can be performed to a target loudspeaker layout (e.g. 5.1 surround) or to headphones while the metadata can be dynamically modified, e.g. in response to a head-tracking device feedback. Other types of rendering may be contemplated. BACKGROUND

[0003] In last years, the generation, recording, representation, coding, transmission, and reproduction of audio is moving towards enhanced, interactive and immersive experience for the listener. The immersive experience can be described e.g. as a state of being deeply engaged or involved in a sound scene while the sounds are coming from all directions. In immersive audio (also called 3D audio), the sound image is reproduced in all 3 dimensions around the listener taking into account a wide range of sound characteristics like timbre, directivity, reverberation, transparency and accuracy of (auditory) spaciousness. Immersive audio is produced for given reproduction systems, i.e. loudspeaker configurations, integrated reproduction systems (sound bars) or headphones. Then interactivity of an audio reproduction system can include e.g. an ability to adjust sound levels, change positions of sounds, or select different languages for the reproduction.

[0004] There are three fundamental approaches (also referred below as audio formats) to achieve an immersive audio experience.

[0005] A first approach is a channel-based audio where multiple spaced microphones are used to capture sounds from different directions while one microphone corresponds to one audio channel in a specific loudspeaker layout. Each recorded channel is supplied to a loudspeaker in a particular location. Examples of channel-based audio comprise, for example, stereo, 5.1 surround, 5.1+4 etc.

[0006] A second approach is a scene-based audio which represents a desired sound field over a localized space as a function of time by a combination of dimensional components. The signals representing the scene-based audio are independent of the audio sources positions while the sound field has to be transformed to a chosen loudspeakers layout at the rendering reproduction system. An example of scene-based audio is ambisonics.

[0007] A third, last immersive audio approach is an object-based audio which represents an auditory scene as a set of individual audio elements (for example singer, drums, guitar) accompanied by information about, for example their position in the audio scene, so that they can be rendered at the reproduction system to their intended locations. This gives an object-based audio a great flexibility and interactivity because each object is kept discrete and can be individually manipulated. The U.S. patent application No. US 2019 / 103118 discloses a multi-stream audio coding. A method includes receiving, at an audio encoder, multiple streams of audio data, where N is the number of the received multi streams. The method includes determining a similarity value for each stream of the multiple streams and comparing the similarity value for each stream of the multiple streams with a threshold. The method also includes identifying, based on the comparison, L (L<N) number of streams to be encoded among the N number of the multiple streams. The method includes encoding the identified L number of streams to generate an encoded bitstream. The U.S. patent application No. US 2015 / 255076 discloses a post-encoding bitrate reduction of multiple object audio. A post-encoding bitrate reduction system and method for generating one more scaled compressed bitstreams from a single encoded plenary file. The plenary file contains multiple audio object files that were encoded separately using a scalable encoding process having fine-grained scalability. Activity in the data frames of the encoded audio object files at a time period are compared with each other to obtain a data frame activity comparison. Bits from an available bitpool are assigned to all of the data frames based on the data frame activity comparison and corresponding hierarchical metadata. The plenary file is scaled down by truncating bits in the data frames to conform to the bit allocation. In some embodiments frame activity is compared to a silence threshold and the data frame contains silence if the frame activity is less than or equal to the threshold and minimal bits are used to represent the silent frame. The U.S. patent application No. US 2016 / 133263 discloses an apparatus and a method for low delay object metadata coding. An apparatus for generating one or more audio channels is provided. The apparatus comprises a metadata decoder for generating one or more reconstructed metadata signals from one or more processed metadata signals depending on a control signal, wherein each of the one or more reconstructed metadata signals indicates information associated with an audio object signal of one or more audio object signals, wherein the metadata decoder is configured to generate the one or more reconstructed metadata signals by determining a plurality of reconstructed metadata samples for each of the one or more reconstructed metadata signals. The apparatus comprises an audio channel generator for generating the one or more audio channels depending on the one or more audio object signals and depending on the one or more reconstructed metadata signals. The metadata decoder is configured to receive a plurality of processed metadata samples of each of the one or more processed metadata signals. The metadata decoder is configured to receive the control signal. The U.S. patent application No. US 2017 / 365262 discloses an audio coding device. An input signal includes a channel-based audio signal and an object-based audio signal, and an audio encoding device includes an audio scene analysis unit configured to determine an audio scene from the input signal and detect audio scene information; a channel-based encoder that encodes the channel-based audio signal output from the audio scene analysis unit; an object-based encoder that encodes the object-based audio signal output from the audio scene analysis unit; and an audio scene encoding unit configured to encode the audio scene information.

[0008] Each of the above described audio formats has its pros and cons. It is thus common that not only one specific format is used in an audio system, but they might be combined in a complex audio system to create an immersive auditory scene. An example can be a system that combines a scene-based or channel-based audio with an object-based audio, e.g. ambisonics with few discrete audio objects.

[0009] The present disclosure presents in the following description a framework to encode and decode object-based audio. Such framework can be a standalone system for object-based audio format coding, or it could form part of a complex immersive codec that may contain coding of other audio formats and / or combination thereof.SUMMARY

[0010] According to a first aspect, the present disclosure provides a system for coding an object-based audio signal according to claim 1.

[0011] The present disclosure also provides a method for coding an object-based audio signal according to claim 20.

[0012] The foregoing and other objects, advantages and features of the system and method for coding an object-based audio signal and the system and method for decoding an object-based audio signal will become more apparent upon reading of the following non-restrictive description of illustrative embodiments thereof, given by way of example only with reference to the accompanying drawings.BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In the appended drawings: Figure 1 is a schematic block diagram illustrating concurrently the system for coding an object-based audio signal and the corresponding method for coding the object-based audio signal; Figure 2 is a diagram showing different scenarios of bit-stream coding of one metadata parameter; Figure 3a is a graph showing values of an absolute coding flag, flag abs , for metadata parameters of three (3) audio objects without using an inter-object metadata coding logic, and Figure 3b is a graph showing values of the absolute coding flag, flag abs , for the metadata parameters of the three (3) audio objects using the inter-object metadata coding logic, wherein arrows indicate frames where the value of several absolute coding flags equal to 1; Figure 4 is a graph illustrating an example of bitrate adaptation for three (3) core-encoders; Figure 5 is a graph illustrating an example of bitrate adaptation based on an ISm (Independent audio stream with metadata) importance logic; Figure 6 is a schematic diagram illustrating the structure of a bit-stream transmitted from the coding system of Figure 1 to the decoding system of Figure 7; Figure 7 is a schematic block diagram illustrating concurrently the system for decoding audio objects in response to audio streams with associated metadata and the corresponding method for decoding the audio objects; and Figure 8 is a simplified block diagram of an example configuration of hardware components implementing the system and method for coding an object-based audio signal and the system and method for decoding the object-based audio signal. DETAILED DESCRIPTION

[0014] The present disclosure provides an example of mechanism for coding the metadata. The present disclosure also provides a mechanism for flexible intra-object and inter-object bitrate adaptation, i.e. a mechanism that distributes the available bitrate as efficiently as possible. In the present disclosure, it is further considered that the bitrate is fixed (constant). However, it is within the scope of the present disclosure to similarly consider an adaptive bitrate, for example (a) in an adaptive bitrate-based codec or (b) as a result of coding a combination of audio formats coded otherwise at a fixed total bitrate.

[0015] There is no description in the present disclosure as to how audio streams are actually coded in a so-called "core-encoder." In general, the core-encoder for coding one audio stream can be an arbitrary mono codec using adaptive bitrate coding. An example is a codec based on the EVS codec as described in Reference [1] with a fluctuating bit-budget that is flexibly and efficiently distributed between modules of the core-encoder, for example as described in Reference [2].1. Framework for coding of audio objects

[0016] As a non-limitative example, the present disclosure considers a framework that supports simultaneous coding of several audio objects (for example up to 16 audio objects) while a fixed constant ISm total bitrate, referred to as ism_total_brate, is considered for coding the audio objects, including the audio streams with their associated metadata. It should be noted that the metadata are not necessarily transmitted for at least some of the audio objects, for example in the case of non-diegetic content. Non-diegetic sounds in movies, TV shows and other videos are sound that the characters cannot hear. Soundtracks are an example of non-diegetic sound, since the audience members are the only ones to hear the music.

[0017] In the case of coding a combination of audio formats in the framework, for example an ambisonics audio format with two (2) audio objects, the constant total codec bitrate, referred to as codec_total_brate, then represents a sum of the ambisonics audio format bitrate (i. e. the bitrate to encode the ambisonics audio format) and the ISm total bitrate ism_total_brate (i.e. the sum of bitrates to code the audio objects, i.e. the audio streams with the associated metadata).

[0018] The present disclosure considers a basic non-limitative example of input metadata consisting of two parameters, namely azimuth and elevation, which are stored per audio frame for each object. In this example, an azimuth range of [-180°, 180°), and an elevation range of [-90°, 90°], is considered. However, it is within the scope of the present disclosure to consider only one or more than two (2) metadata parameters.2. Object-based coding

[0019] Figure 1 is a schematic block diagram illustrating concurrently the system 100, comprising several processing blocks, for coding an object-based audio signal and the corresponding method 150 for coding the object-based audio signal.2.1 Input buffering

[0020] Referring to Figure 1, the method 150 for coding the object-based audio signal comprises an operation of input buffering 151. To perform the operation 151 of input buffering, the system 100 for coding the object-based audio signal comprises an input buffer 101.

[0021] The input buffer 101 buffers a number N of input audio objects 102, i.e. a number N of audio streams with the associated respective N metadata. The N input audio objects 102, including the N audio streams and the N metadata associated to each of these N audio streams are buffered for one frame, for example a 20 ms long frame. As well known in the art of sound signal processing, the sound signal is sampled at a given sampling frequency and processed by successive blocks of these samples called "frames" each divided into a number of "sub-frames."2.2 Audio streams analysis and front pre-processing

[0022] Still referring to Figure 1, the method 150 for coding the object-based audio signal comprises an operation of analysis and front pre-processing 153 of the N audio streams. To perform the operation 153, the system 100 for coding the object-based audio signal comprises an audio stream processor 103 to analyze and front pre-process, for example in parallel, the buffered N audio streams transmitted from the input buffer 101 to the audio stream processor 103 through a number N of transport channels 104, respectively.

[0023] The analysis and front pre-processing operation 153 performed by the audio stream processor 103 may comprise, for example, at least one of the following sub-operations: time-domain transient detection, spectral analysis, long-term prediction analysis, pitch tracking and voicing analysis, voice / sound activity detection (VAD / SAD), bandwidth detection, noise estimation and signal classification (which may include in a non-limitative embodiment (a) core-encoder selection between, for example, ACELP core-encoder, TCX core-encoder, HQ core-encoder, etc., (b) signal type classification between, for example, inactive core-encoder type, unvoiced core-encoder type, voiced core-encoder type, generic core-encoder type, transition core-encoder type, and audio core-encoder type, etc., (c) speech / music classification, etc.). Information obtained from the analysis and front pre-processing operation 153 is supplied to a configuration and decision processor 106 through la line 121. Examples of the foregoing sub-operations are described in Reference [1] in relation to the EVS codec and, therefore, will not be further described in the present disclosure.2.3 Metadata analysis, quantization and coding

[0024] The method 150 of Figure 1, for coding the object-based audio signal comprises an operation of metadata analysis, quantization and coding 155. To perform the operation 155, the system 100 for coding the object-based audio signal comprises a metadata processor 105.2.3.1 Metadata analysis

[0025] Signal classification information 120 (for example VAD or localVAD flag as used in the EVS codec (See Reference [1]) from the audio stream processor 103 is supplied to the metadata processor 105. The metadata processor 105 comprises an analyzer (not shown) of the metadata of each of the N audio objects to determine whether the current frame is inactive (for example VAD = 0) or active (for example VAC ≠ 0) with respect to this particular audio object. In inactive frames, no metadata is coded by the metadata processor 105 relative of that object. In active frames, the metadata are quantized and coded for this audio object using a variable bitrate. More details about metadata quantization and coding will be provided in the following Sections 2.3.2 and 2.3.3.2.3.2 Metadata quantization

[0026] The metadata processor 105 of Figure 1 quantizes and codes the metadata of the N audio objects, in the described non-restrictive illustrative embodiments, sequentially in a loop while a certain dependency can be employed between quantization of audio objects and the metadata parameters of these audio objects.

[0027] As indicated herein above, in the present disclosure, two metadata parameters, azimuth and elevation (as included in the N input metadata), are considered. As a non-limitative example, the metadata processor 105 comprises a quantizer (not shown) of the following metadata parameter indexes using the following example resolution to reduce the number of bits being used: Azimuth parameter: A 12-bit azimuth parameter index from a file of the input metadata is quantized to B az -bit index (for example B az = 7). Giving the minimum and maximum azimuth limits (-180 and +180°), a quantization step for a (B az = 7)-bit uniform scalar quantizer is 2.835°. Elevation parameter: A 12-bit elevation parameter index from the input metadata file is quantized to B el -bit index (for example B el = 6). Giving the minimum and maximum elevation limits (-90 0< and +90 0< ), a quantization step for a (B el = 6)-bit uniform scalar quantizer is 2.857 0< .

[0028] A total metadata bit-budget for coding the N metadata and a total number quantization bits for quantizing the metadata parameter indexes (i.e. the quantization index granularity and thus the resolution) may be made dependent on the bitrate(s) codec_total_brate, ism_total_brate and / or element_brate (the latter resulting from a sum of a metadata bit-budget and / or a core-encoder bit-budget related to one audio object).

[0029] The azimuth and elevation parameters can be represented as one parameter, for example by a point on a sphere. In such a case, it is within the scope of the present disclosure to implement different metadata including two or more parameters.2.3.3 Metadata coding

[0030] Both azimuth and elevation indexes, once quantized, can be coded by a metadata encoder (not shown) of the metadata processor 105 using either absolute or differential coding. As known, absolute coding means that a current value of a parameter is coded. Differential coding means that a difference between a current value and a previous value of a parameter is coded. As the indexes of the azimuth and elevation parameters usually evolve smoothly (i.e. a change in azimuth or elevation position can be considered as continuous and smooth), differential coding is used by default. However, absolute coding may be used, for example in the following instances: There is too large a difference between current and previous values of the parameter index which would result in a higher or equal number of bits for using differential coding compared to using absolute coding (may happen exceptionally); No metadata were coded and sent in the previous frame; There were too many consecutive frames with differential coding. In order to control decoding in a noisy channel (Bad Frame Indicator, BFI = 1). For example, the metadata encoder codes the metadata parameter indexes using absolute coding if a number of consecutive frames which are coded using differential is higher that a maximum number of consecutive frames coded using different coding. The latter maximum number of consecutive frames is set to β. In a non-restrictive illustrative example, β = 10 frames.

[0031] The metadata encoder produces a 1-bit absolute coding flag, flag abs , to distinguish between absolute and differential coding.

[0032] In the case of absolute coding, the coding flag, flag abs , is set to 1, and is followed by the B az -bit (or B el -bit) index coded using absolute coding, where B az and B el refer to the above mentioned indexes of the azimuth and elevation parameters to be coded, respectively.

[0033] In the case of differential coding, the 1-bit coding flag, flag abs , is set to 0 and is followed by a 1-bit zero coding flag, flag zero , signaling a difference Δ between the B az -bit indexes (respectively the B el -bit indices) in the current and previous frames equal to 0. If the difference Δ is not equal to 0, the metadata encoder continues coding by producing a 1-bit sign flag, flag sign , followed by a difference index, of which the number of bits is adaptive, in a form of, for example, a unary code indicative of the value of the difference Δ.

[0034] Figure 2 is a diagram showing different scenarios of bit-stream coding of one metadata parameter.

[0035] Referring to Figure 2, it is noted that not all metadata parameters are always transmitted in every frame. Some might be transmitted only in every y th< frame, some are not sent at all for example when they do not evolve, they are not important or the available bit-budget is low. Referring to Figure 2, for example: in the case of absolute coding (first line of Figure 2), the absolute coding flag, flag abs , and the B az -bit index (respectively the B el -bit index) are transmitted; in the case of differential coding with the difference Δ between the B az -bit indexes (respectively the B el -bit indexes) in the current and previous frames equal to 0 (second line of Figure 2), the absolute coding flag, flag abs =0, and the zero coding flag, flag zero =1 are transmitted; in the case of differential coding with a positive difference Δ between the B az -bit index (respectively the B el -bit indexes) in the current and previous frames (third line of Figure 2), the absolute coding flag, flag abs =0, the zero coding flag, flag zero =0, the sign flag, flag sign =0, and the difference index (1 to (B az -3)-bits index (respectively 1 to (B el -3)-bits index)) are transmitted; and in the case of differential coding with a negative difference Δ between the B az -bit indexes (respectively the B el -bit indexes) in the current and previous frames (last line of Figure 2), the absolute coding flag, flag abs =0, the zero coding flag, flag zero =0, the sign flag, flag sign =1, and the difference index (1 to (B az -3)-bits index (respectively 1 to (B el -3)-bits index)) are transmitted. 2.3.3.1 Intra-object metadata coding logic

[0036] The logic used to set absolute or differential coding may be further extended by an intra-object metadata coding logic. Specifically, in order to limit a range of metadata coding bit-budget fluctuation between frames and thus to avoid too low a bit-budget left for the core-encoders 109, the metadata encoder limits absolute coding in a given frame to one, or generally to a number as low as possible of, metadata parameters.

[0037] In the non-limitative example of azimuth and elevation metadata parameter coding, the metadata encoder uses a logic that avoids absolute coding of the elevation index in a given frame if the azimuth index was already coded using absolute coding in the same frame. In other words, the azimuth and elevation parameters of one audio object are (practically) never both coded using absolute coding in a same frame. As a consequence, the absolute coding flag, flag abs.ele , for the elevation parameter is not transmitted in the audio object bit-stream if the absolute coding flag, flag abs.azi , for the azimuth parameter is equal to 1.

[0038] It is also within the scope of the present disclosure to make the intra-object metadata coding logic bitrate dependent. For example, both the absolute coding flag, flag abs.ele , for the elevation parameter and the absolute coding flag, flag abs.azi , for the azimuth parameter can be transmitted in a same frame is the bitrate is sufficiently large.2.3.3.2 Inter-object metadata coding logic

[0039] The metadata encoder may apply a similar logic to metadata coding of different audio objects. The implemented inter-object metadata coding logic minimizes the number of metadata parameters of different audio objects coded using absolute coding in a current frame. This is achieved by the metadata encoder mainly by controlling frame counters of metadata parameters coded using absolute coding chosen from robustness purposes and represented by the parameter β. As a non-limitative example, a scenario where the metadata parameters of the audio objects evolve slowly and smoothly is considered. In order to control decoding in a noisy channel where indexes are coded using absolute coding every β frames, the azimuth B az -bit index of audio object #1 is coded using absolute coding in frame M, the elevation B el -bit index of audio object #1 is coded using absolute coding in frame M+1, the azimuth B az -bit index of audio object #2 is encoded using absolute coding in frame M+2, the elevation B el -bit index of object #2 is coded using absolute coding in frame M+3, etc.

[0040] Figure 3a is a graph showing values of the absolute coding flag, flag abs , for metadata parameters of three (3) audio objects without using the inter-object metadata coding logic, and Figure 3b is a graph showing values of the absolute coding flag, flag abs , for the metadata parameters of the three (3) audio objects using the inter-object metadata coding logic. In Figure 3a, the arrows indicate frames where the value of several absolute coding flags is equal to 1.

[0041] More specifically, Figure 3a shows the values of the absolute coding flag, flag abs , for two metadata parameters (azimuth and elevation in this particular example) for the audio objects without using the inter-object metadata coding logic, while Figure 3b shows the same values but with the inter-object metadata coding logic implemented. The graphs of Figures 3a and 3b correspond to (from top to bottom): audio stream of audio object #1; audio stream of audio object #2; audio stream of audio object #3, absolute coding flag, flag abs,azi , for the azimuth parameter of audio object #1; absolute coding flag, flag abs,ele , for the elevation parameter of audio object #1; absolute coding flag, flag abs,azi , for the azimuth parameter of audio object #2; absolute coding flag, flag abs,ele , for the elevation parameter of audio object #2; absolute coding flag, flag abs,azi , for the azimuth parameter of audio object #3; and absolute coding flag, flag abs,ele , for the elevation parameter of audio object #3.

[0042] It can be seen from Figure 3a that several flag abs may have a value equal to 1 (see the arrows) in a same frame when the inter-object metadata coding logic is not used. In contrast, Figure 3b shows that only one absolute flag, flag abs , may have a value equal to 1 in a given frame when the inter-object metadata coding logic is used.

[0043] The inter-object metadata coding logic may also be made bitrate dependent. In this case, for example, more that one absolute flag, flag abs , may have a value equal to 1 in a given frame even when the inter-object metadata coding logic is used, if the bitrate is sufficiently large.

[0044] A technical advantage of the inter-object metadata coding logic and the intra-object metadata coding logic is to limit a range of fluctuation of the metadata coding bit-budget between frames. Another technical advantage is to increase robustness of the codec in a noisy channel; when a frame is lost, then only a limited number of metadata parameters from the audio objects coded using absolute coding is lost. Consequently, any error propagated from a lost frame affects only a small number of metadata parameters across the audio objects and thus does not affect the whole audio scene (or several different channels).

[0045] A global technical advantage of analyzing, quantizing and coding the metadata separately from the audio streams is, as described hereinabove, to enable processing specially adapted to the metadata and more efficient in terms of metadata coding bitrate, metadata coding bit-budget fluctuation, robustness in noisy channel, and error propagation due to lost frames.

[0046] The quantized and coded metadata 112 from the metadata processor 105 are supplied to a multiplexer 110 for insertion into an output bit-stream 111 transmitted to a distant decoder 700 (Figure 7).

[0047] Once the metadata of the N audio objects are analyzed, quantized and encoded, information 107 from the metadata processor 105 about the bit-budget for the coding of the metadata per audio object is supplied to a configuration and decision processor 106 (bit-budget allocator) described in more detail in the following section 2.4. When the configuration and bitrate distribution between the audio streams is completed in processor 106 (bit-budget allocator), the coding continues with further pre-processing 158 to be described later. Finally, the N audio streams are encoded using an encoder comprising, for example, N fluctuating bitrate core-encoders 109, such as mono core-encoders.2.4 Bitrates per channel configuration and decision

[0048] The method 150 of Figure 1, for coding the object-based audio signal comprises an operation 156 of configuration and decision about bitrates per transport channel 104. To perform the operation 156, the system 100 for coding the object-based audio signal comprises the configuration and decision processor 106 forming a bit-budget allocator.

[0049] The configuration and decision processor 106 (herein after bit-budget allocator 106) uses a bitrate adaptation algorithm to distribute the available bit-budget for core-encoding the N audio streams in the N transport channels 104.

[0050] The bitrate adaptation algorithm of the configuration and decision operation 156 comprises the following sub-operations 1-6 performed by the bit-budget allocator 106: 1. The ISm total bit-budget, bits ism , per frame is calculated from the ISm total bitrate ism_total_brate (or the codec total bitrate codec_total_brate if only audio objects are coded) using, for example, the following relation: bits ism = ism _ total _ brate 50 The denominator, 50, corresponds to the number of frames per second, assuming 20-ms long frames. The value 50 would be different if the size of the frame is different from 20 ms. 2. The above defined element bitrate element_brate (resulting from a sum of the metadata bit-budget and core-encoder bit-budget related to one audio object) defined for N audio objects is supposed to be constant during a session at a given codec total bitrate, and about the same for the N audio objects. A "session" is defined for example as a phone call or an off-line compression of an audio file. The corresponding element bit-budget, bits element , is computed for the audio streams objects n = 0, ..., N-1 using, for example, the following relation: bits element n = bits ism N where x indicates the largest integer smaller than or equal to x. In order to spend all available ISm total bit-budget bits ism the element bit-budget bits element of, for example, the last audio object is eventually adjusted using the following relation: bits element N − 1 = bits ism N + bits ism mod N where "mod" indicates a remainder modulo operation. Finally, the element bit-budget bits element of the N audio objects is used to set the value element_brate for the ausio objects n = 0, ..., N-1 using, for example, the following relation: element _ brate n = bits element n * 50 where the number 50, as already mentioned, corresponds to the number of frames per second, assuming 20-ms long frames. 3. The metadata bit-budget bits meta , per frame, of the N audio objects is summed, using the following relation: bits meta _ all = ∑ n = 0 N − 1 bits meta n and the resulting value bits metal_all is added to an ISm common signaling bit-budget, bits Ism_signalling , resulting in the codec side bit-budget: bits side = bits meta _ all + bits ISm _ signalling 4. The codec side bit-budget, bits side , per frame, is split equally between the N audio objects and used to compute the core-encoder bit-budget, bits CoreCoder , for each of the N audio streams using, for example, the following relation: bits CoreCoder n = bits element n − bits side N while the core-encoder bit-budget of, for example, the last audio stream may eventually be adjusted to spend all the available core-encoding bit-budget using, for example, the following relation: biys CoreCoder N − 1 = bits element N − 1 − bits side N + bits side mod N The corresponding total bitrate, total_brate, i.e. the bitrate to code one audio stream, in a core-encoder, is then obtained for n = 0, ..., N-1 using, for example, the following relation: total _ brate n = bits CoreCoder n * 50 where the number 50, again, corresponds to the number of frames per second, assuming 20-ms long frames. 5. The total bitrate, total_brate, in inactive frames (or in frames with very low energy or otherwise without meaningful content) may be lowered and set to a constant value in the related audio streams. The so saved bit-budget is then redistributed equally between the audio streams with active content in the frame. Such redistribution of bit-budget will be further described in the following section 2.4.1. 6. The total bitrate, total_brate, in audio streams (with active content) in active frames is further adjusted between these audio streams based on an ISm importance classification. Such adjustment of bitrate will be further described in the following section 2.4.2.

[0051] When the audio streams are all in an inactive segment (or are without meaningful content), the above last two sub-operations 5 and 6 may be skipped. Accordingly, the bitrate adaptation algorithms described in following sections 2.4.1 and 2.4.2 are employed when at least one audio stream has active content.2.4.1 Bitrate adaptation based on signal activity

[0052] In inactive frames (VAD = 0), the total bitrate, total_brate, is lowered and the saved bit-budget is redistributed, for example equally between the audio streams in active frames (VAD ≠ 0). The assumption is that waveform coding of an audio stream in frames which are classified as inactive is not required; the audio object may be muted. The logic, used in every frame, can be expressed by the following sub-operations 1-3: 1. For a particular frame, set a lower core-encoder bit-budget to every audio stream n with inactive content: bits CoreCoder ′ n = B VAD 0 ∀ n with VAD = 0 where B VAD0 is a lower, constant core-encoder bit-budget to be set in inactive frames; for example B VAD0 = 140 (corresponding to 7 kbps for a 20 ms frame) or B VAD0 = 49 (corresponding to 2.45 kbps for a 20 ms frame). 2. Next, the saved bit-budget is computed using, for example, the following relation: bits diff = ∑ n = 0 N − 1 bits CoreCoder ′ n − bits CoreCoder n 3. Finally, the saved bit-budget is redistributed, for example equally between the core-encoder bit-budgets of the audio streams with active content in a given frame using the following relation: bits CoreCoder ′ n = bits CoreCoder n + bits diff N VAD 1 ∀n with VAD=1 where N VAD1 is the number of audio streams with active content. The core-encoder bit-budget of the first audio stream with active content is eventually increased using, for example, the following relation: bits CoreCoder ′ n = bits CoreCoder n + bits diff N VAD 1 + bits diff mod N VAD 1 , ∀n Afirst VAD=1 stream The corresponding core-encoder total bitrate, total_brate, is finally obtained for each audio stream n = 0, ..., N-1 as follows: total _ brate ′ n = bits CoreCoder ′ n * 50

[0053] Figure 4 is a graph illustrating an example of bitrate adaptation for three (3) core-encoders. Specifically, In Figure 4, the first line shows the core-encoder total bitrate, total_brate, for audio stream #1, the second line shows the core-encoder total bitrate, total_brate, for audio stream #2, the third line shows the core-encoder total bitrate, total_brate, for audio stream #3, line 4 is the audio stream #1, line 5 is the audio stream #2, and line 4 is the audio stream #3.

[0054] In the example of Figure 4, the adaptation of the total bitrate, total_brate, for the three (3) core-encoder is based on VAD activity (active / inactive frames). As can be seen from Figure 4, most of the time there is a small fluctuation of the core-encoder total bitrate, total_brate, as a result of the fluctuating side bit-budget bits side . Then, there are infrequent substantial changes of the core-encoder total bitrate, total_brate, as a result of the VAD activity.

[0055] For example, referring to Figure 4, instance A) corresponds to a frame where the audio stream #1 VAD activity changes from 1 (active) to 0 (inactive). According to the logic, a minimum core-encoder total bitrate, total_brate, is assigned to audio object #1 while the core-encoder total bitrates, total_brate, for active audio objects #2 and #3 are increased. Instance B) corresponds to a frame where the VAD activity of the audio stream #3 changes from 1 (active) to 0 (inactive) while the VAD activity of the audio stream #1 remains to 0. Accordingly to the logic, a minimum core-encoder total bitrate, total_brate, is assigned to audio streams #1 and #3 while the core-encoder total bitrate, total_brate, of the active audio stream #2 is further increased.

[0056] The above logic of section 2.4.1 can be made dependent from the total bitrate ism_total_brate. For example, the bit-budget B VAD0 in the above sub-operation 1 can be set higher for a higher total bitrate ism_total_brate, and lower for a lower total bitrate ism_total_brate.2.4.2 Bitrate adaptation based on ISm importance

[0057] The logic described in previous section 2.4.1 results in about a same core-encoder bitrate in every audio stream with active content (VAD = 1) in a given frame. However, it may be beneficial to introduce an inter-object core-encoder bitrate adaptation based on a classification of ISm importance (or, more generally, on a metric indicative of how critical coding of a particular audio object in a current frame to obtain a given (decent) quality of the decoded synthesis is).

[0058] The classification of ISm importance can be based on several parameters and / or combination of parameters, for example core-encoder type (coder_type), FEC (Forward Error Correction), sound signal classification (class), speech / music classification decision, and / or SNR (Signal-to-Noise Ratio) estimate from the open-loop ACELP / TCX (Algebraic Code-Excited Linear Prediction / Transform-Coded eXcitation) core decision module (snr_celp, snr_tcx) as described in Reference [1]. Other parameters can possibly be used for determining the classification of ISm importance.

[0059] In a non-restrictive example, a simple classification of ISm importance is based on the core-encoder type as defined in Reference [1] is implemented. For that purpose, the bit-budget allocator 106 of Figure 1 comprises a classifier (not shown) for rating the importance of a particular ISm stream. As a result, four (4) distinct ISm importance classes, class ISm , are defined: No metadata class, ISM_NO_META: frames without metadata coding, e.g. inactive frames with VAD = 0; Low importance class, ISM_LOW_IMP: frames where coder_type = UNVOICED or INACTIVE; Medium importance class, ISM_MEDIUM_IMP: frames where coder_type = VOICED; High importance class ISM_HIGH_IMP: frames where coder_type = GENERIC.

[0060] The ISm importance class is then used by the bit-budget allocator 106, in the bitrate adaptation algorithm (See above Section 2.4, sub-operation 6) to assign a higher bit-budget to audio streams with a higher ISm importance and a lower bit-budget to audio streams with a lower ISm importance. Thus for every audio stream n, n = 0,...,N-1, the following bitrate adaptation algorithm is used by the bit-budget allocator 106: 1. In frames classified as class ISm = ISM_NO_META, the constant low bitrate B VAD0 is assigned. 2. In frames classified as class ISm = ISM_LOW_IMP, the total bitrate, total_brate, is lowered for example as: total _ brate new n = max α low * total _ brate n , B low where the constant α low is set to a value lower than 1.0, for example 0.6. Then the constant B low represents a minimum bitrate threshold supported by the codec for a particular configuration, which may be dependent upon, for example, the internal sampling rate of the codec, the coded audio bandwidth, etc. (See Reference [1] for more detail about these values). 3. In frames classified as class ISm = ISM_MEDIUM_IMP: the core-encoder total bitrate, total_brate, is lowered for example as total _ brate new n = max α med * total _ brate n , B low where the constant α med is set to a value lower than 1.0 but higher than α low , for example to 0.8. 4. In frames classified as class ISm = ISM_HIGH_IMP, no bitrate adaptation is used; 5. Finally, the saved bit-budget (a sum of differences between the old (total_brate) and new (total_brate new ) total bitrates) is redistributed equally between the audio streams with active content in the frame. The same bit-budget redistribution logic as described in section 2.4.1, sub-operations 2 and 3, may be used.

[0061] Figure 5 is a graph illustrating an example of bitrate adaptation based on ISm importance logic. From top to bottom, the graph of Figure 5 illustrates, in time: An active speech segment of the audio stream for audio object #1; An active speech segment of the audio stream for audio object #2; The total bitrate, total_brate, of the audio stream for audio object #1 without using the bitrate adaptation algorithm; The total bitrate, total_brate, of the audio stream for audio object #2 without using the bitrate adaptation algorithm; The total bitrate, total_brate, of the audio stream for audio object #1 when the bitrate adaptation algorithm is used; and The total bitrate, total_brate, of the audio stream for audio object #2 when the bitrate adaptation algorithm is used.

[0062] In the non-limitative example of Figure 5, with two audio objects (N=2) and a fixed constant total bitrate, ism_total_brate, equal to 48 kbps, the core-encoder total bitrate, total_brate, in active frames of audio object #1 fluctuates between 23.45 kbps and 23.65 kbps when the bitrate adaptation algorithm is not used while it fluctuates between 19.15 kbps and 28.05 kbps when the bitrate adaptation algorithm is used. Similarly, the core-encoder total bitrate, total_brate, in active frames of audio object #2 fluctuates between 23.40 kbps and 23.65 kbps without using the bitrate adaptation algorithm and between 19.10 kbps and 28.05 kbps with the bitrate adaptation algorithm. A better, more efficient distribution of the available bit-budget between the audio streams is thereby obtained.2.5 Pre-processing

[0063] Referring to Figure 1, the method 150 for coding the object-based audio signal comprises an operation of pre-processing 158 of the N audio streams conveyed through the N transport channels 104 from the configuration and decision processor 106 (bit-budget allocator). To perform the operation 158, the system 100 for coding the object-based audio signal comprises a pre-processor 108.

[0064] Once the configuration and bitrate distribution between the N audio streams is completed by the configuration and decision processor 106 (bit-budget allocator), the pre-processor 108 performs sequential further pre-processing 158 on each of the N audio streams. Such pre-processing 158 may comprise, for example, further signal classification, further core-encoder selection (for example selection between ACELP core, TCX core, and HQ core), other resampling at a different internal sampling frequency F s adapted to the bitrate to be used for core-encoding, etc. Examples of such pre-processing can be found, for example, in Reference [1] in relation to the EVS codec and, therefore, will not be further described in the present disclosure.2.6 Core-encoding

[0065] Referring to Figure 1, the method 150 for coding the object-based audio signal comprises an operation of core-encoding 159. To perform the operation 159, the system 100 for coding the object-based audio signal comprises the above mentioned encoder of the N audio streams including, for example, a number N of core-encoders 109 to respectively code the N audio streams conveyed through the N transport channels 104 from the pre-processor 108.

[0066] Specifically, the N audio streams are encoded using N fluctuating bitrate core-encoders 109, for example mono core-encoders. The bitrate used by each of the N core-encoders is the bitrate selected by the configuration and decision processor 106 (bit-budget allocator) for the corresponding audio stream. For example, core-encoders as described in Reference [1] can be used as core-encoders 109.3.0 Bit-stream structure

[0067] Referring to Figure 1, the method 150 for coding the object-based audio signal comprises an operation of multiplexing 160. To perform the operation 160, the system 100 for coding the object-based audio signal comprises a multiplexer 110.

[0068] Figure 6 is a schematic diagram illustrating, for a frame, the structure of the bit-stream 111 produced by the multiplexer 110 and transmitted from the coding system 100 of Figure 1 to the decoding system 700 of Figure 7. Regardless whether metadata are present and transmitted or not, the structure of the bit-stream 111 may be structured as illustrated in Figure 6.

[0069] Referring to Figure 6, the multiplexer 110 writes the indices of the N audio streams from the beginning of the bit-stream 111 while the indices of ISm common signaling 113 from the configuration and decision processor 106 (bit-budget allocator) and metadata 112 from the metadata processor 105 are written from the end of the bit-stream 111.3.1 ISm common signaling

[0070] The multiplexer writes the ISm common signaling 113 from the end of the bit-stream 111. The ISm common signaling is produced by the configuration and decision processor 106 (bit-budget allocator) and comprises a variable number of bits representing: (a) a number N of audio objects: the signaling for the number N of coded audio objects present in the bit-stream 111 is in the form of, for example, a unary code with a stop bit (e.g. for N = 3 audio objects, the first 3 bits of the ISm common signaling would be "110"). (b) a metadata presence flag, flag meta : The flag, flag meta , is present when the bitrate adaptation based on signal activity as described in section 2.4.1 is used and comprises one bit per audio object to indicate whether metadata for that particular audio object are present (flag meta = 1) or not (flag meta = 0) in the bit-stream 111, or (c) the ISm importance class: this signaling is present when the bitrate adaptation based on the ISM importance as described in section 2.4.2 is used and comprises two bits per audio object to indicate the ISm importance class, class Ism (ISM_NO_META, ISM_LOW_IMP, ISM_MEDIUM_IMP, and ISM_HIGH_IMP), as defined in section 2.4.2. (d) an ISm VAD flag, flag VAD : the ISm VAD flag is transmitted when flag meta = 0, respectively class Ism = ISM_NO_META, and distinguishes between the following two cases: 1) input metadata are not present or metadata are not coded so that the audio stream needs to be coded by an active coding mode (flag VAD = 1); and 2) input metadata are present and transmitted so that the audio stream can be coded by an inactive coding mode (flag VAD = 0). 3.2 Coded metadata payload

[0071] The multiplexer 110 is supplied with the coded metadata 112 from the metadata processor 105 and writes the metadata payload sequentially from the end of the bit-stream for the audio objects for which the metadata are coded (flag meta = 1, respectively class Ism ≠ ISM_NO_META) in the current frame. The metadata bit-budget for each audio object is not constant but rather inter-object and inter-frame adaptive. Different metadata format scenarios are shown in Figure 2.

[0072] In the case that metadata are not present or are not transmitted for at least some of the N audio objects, the metadata flag is set to 0, i.e. flag meta = 0, respectively class Ism = ISM_NO_META, for these audio objects. Then, no metadata indices are sent in relation to those audio objects, i.e. bits meta [n] =0.3.3 Audio streams payload

[0073] The multiplexer 110 receives the N audio streams 114 coded by the N core encoders 109 through the N transport channels 104, and writes the audio streams payload sequentially for the N audio streams in chronological order from the beginning of the bit-stream 111 (See Figure 6). The respective bit-budgets of the N audio streams are fluctuating as a result of the bitrate adaptation algorithm described in section 2.4.4.0 Decoding of audio objects

[0074] Figure 7 is a schematic block diagram illustrating concurrently the system 700 for decoding audio objects in response to audio streams with associated metadata and the corresponding method 750 for decoding the audio objects.4.1 Demultiplexing

[0075] Referring to Figure 7, the method 750 for decoding audio objects in response to audio streams with associated metadata comprises an operation of demultiplexing 755. To perform the operation 755, the system 700 for decoding audio objects in response to audio streams with associated metadata comprises a demultiplexer 705.

[0076] The demultiplexer receive a bit-stream 701 transmitted from the coding system 100 of Figure 1 to the decoding system 700 of Figure 7. Specifically, the bit-stream 701 of Figure 7 corresponds to the bit-stream 111 of Figure 1.

[0077] The demultiplexer 110 extracts from the bit-stream 701 (a) the coded N audio streams 114, (b) the coded metadata 112 for the N audio objects, and (c) the ISm common signaling 113 read from the end of the received bit-stream 701.4.2 Metadata decoding and dequantization

[0078] Referring to Figure 7, the method 750 for decoding audio objects in response to audio streams with associated metadata comprises an operation 756 of metadata decoding and dequantization. To perform the operation 756, the system 700 for decoding audio objects in response to audio streams with associated metadata comprises a metadata decoding and dequantization processor 706.

[0079] The metadata decoding and dequantization processor 706 is supplied with the coded metadata 112 for the transmitted audio objects, the ISm common signaling 113, and an output set-up 709 to decode and dequantize the metadata for the audio streams / objects with active contents. The output set-up 709 is a command line parameter about the number M of decoded audio objects / transport channels and / or audio formats, which can be equal to or different from the number N of coded audio objects / transport channels. The metadata decoding and dequantization processor 706 produces decoded metadata 704 for the M audio objects / transport channels, and supplies information about the respective bit-budgets for the M decoded metadata on line 708. Obviously, the decoding and dequantization performed by the processor 706 is the inverse of the quantization and coding performed by the metadata processor 105 of Figure 1.4.3 Configuration and decision about bitrates

[0080] Referring to Figure 7, the method 750 for decoding audio objects in response to audio streams with associated metadata comprises an operation 757 of configuration and decision about bitrates per channel. To perform the operation 757, the system 700 for decoding audio objects in response to audio streams with associated metadata comprises a configuration and decision processor 707 (bit-budget allocator).

[0081] The bit-budget allocator 707 receives (a) the information about the respective bit-budgets for the M decoded metadata on line 708 and (b) the ISm importance class, class ISm , from the common signaling 113, and determines the core-decoder bitrates per audio stream, total_brate[n]. The bit-budget allocator 707 uses the same procedure as in the bit-budget allocator 106 of Figure 1 to determine the core-decoder bitrates (see section 2.4).4.4 Core-decoding

[0082] Referring to Figure 7, the method 750 for decoding audio objects in response to audio streams with associated metadata comprises an operation of core-decoding 760. To perform the operation 760, the system 700 for decoding audio objects in response to audio streams with associated metadata comprises a decoder of the N audio streams 114 including a number N of core-decoders 710, for example N fluctuating bitrate core-decoders.

[0083] The N audio streams 114 from the demultiplexer 705 are decoded, for example sequentially decoded in the number N of fluctuating bitrate core decoders 710 at their respective core-decoder bitrates as determined by the bit-budget allocator 707. When the number of decoded audio objects, M, as requested by the output set-up 709 is lower than the number of transport channels, i.e M < N, a lower number of core-decoders are used. Similarly, not all metadata payloads may be decoded in such a case.

[0084] In response to the N audio streams 114 from the demultiplexer 705, the core-decoder bitrates as determined by the bit-budget allocator 707, and the output set-up 709, the core-decoders 710 produces a number M of decoded audio streams 703 on respective M transport channels.5.0 Audio channel rendering

[0085] In an operation of audio channel rendering 761, a renderer 711 of audio objects transforms the M decoded metadata 704 and the M decoded audio streams 703 into a number of output audio channels 702, taking into consideration an output set-up 712 indicative of the number and contents of output audio channels to be produced. Again, the number of output audio channels 702 may be equal to or different from the number M.

[0086] The renderer 761 may be designed in a variety of different structures to obtain the desired output audio channels. For that reason, the renderer will not be further described in the present disclosure.6.0 Source code

[0087] According to a non-limitative illustrative embodiment, the system and method for coding an object-based audio signal as disclosed in the foregoing description may be implemented by the following source code (expressed in C-code) given herein below as additional disclosure. 7.0 Hardware implementation

[0088] Figure 8 is a simplified block diagram of an example configuration of hardware components forming the above described coding and decoding systems and methods.

[0089] Each of the coding and decoding systems may be implemented as a part of a mobile terminal, as a part of a portable media player, or in any similar device. Each of the coding and decoding systems (identified as 1200 in Figure 8) comprises an input 1202, an output 1204, a processor 1206 and a memory 1208.

[0090] The input 1202 is configured to receive the input signal(s), e.g. the N audio objects 102 (N audio streams with the corresponding N metadata) of Figure 1 or the bit-stream 701 of Figure 7, in digital or analog form. The output 1204 is configured to supply the output signal(s), e.g. the bit-stream 111 of Figure 1 or the M decoded audio channels 703 and the M decoded metadata 704 of Figure 7. The input 1202 and the output 1204 may be implemented in a common module, for example a serial input / output device.

[0091] The processor 1206 is operatively connected to the input 1202, to the output 1204, and to the memory 1208. The processor 1206 is realized as one or more processors for executing code instructions in support of the functions of the various processors and other modules of Figures 1 and 7.

[0092] The memory 1208 may comprise a non-transient memory for storing code instructions executable by the processor(s) 1206, specifically, a processor-readable memory comprising non-transitory instructions that, when executed, cause a processor(s) to implement the operations and processors / modules of the coding and decoding systems and methods as described in the present disclosure. The memory 1208 may also comprise a random access memory or buffer(s) to store intermediate processing data from the various functions performed by the processor(s) 1206.

[0093] Those of ordinary skill in the art will realize that the description of the coding and decoding systems and methods are illustrative only and are not intended to be in any way limiting. Other embodiments will readily suggest themselves to such persons with ordinary skill in the art having the benefit of the present disclosure. Furthermore, the disclosed coding and decoding systems and methods may be customized to offer valuable solutions to existing needs and problems of encoding and decoding sound.

[0094] In the interest of clarity, not all of the routine features of the implementations of the coding and decoding systems and methods are shown and described. It will, of course, be appreciated that in the development of any such actual implementation of the coding and decoding systems and methods, numerous implementation-specific decisions may need to be made in order to achieve the developer's specific goals, such as compliance with application-, system-, network- and business-related constraints, and that these specific goals will vary from one implementation to another and from one developer to another. Moreover, it will be appreciated that a development effort might be complex and time-consuming, but would nevertheless be a routine undertaking of engineering for those of ordinary skill in the field of sound processing having the benefit of the present disclosure.

[0095] In accordance with the present disclosure, the processors / modules, processing operations, and / or data structures described herein may be implemented using various types of operating systems, computing platforms, network devices, computer programs, and / or general purpose machines. In addition, those of ordinary skill in the art will recognize that devices of a less general purpose nature, such as hardwired devices, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), or the like, may also be used. Where a method comprising a series of operations and sub-operations is implemented by a processor, computer or a machine and those operations and sub-operations may be stored as a series of non-transitory code instructions readable by the processor, computer or machine, they may be stored on a tangible and / or non-transient medium.

[0096] The coding and decoding systems and methods as described herein may use software, firmware, hardware, or any combination(s) of software, firmware, or hardware suitable for the purposes described herein.

[0097] In the coding and decoding systems and methods as described herein, the various operations and sub-operations may be performed in various orders and some of the operations and sub-operations may be optional.8.0 References

[0098] The following references are referred to: [1] 3GPP Spec. TS 26.445: "Codec for Enhanced Voice Services (EVS). Detailed Algorithmic Description," v.12.0.0, Sep. 2014. [2] V. Eksler, "Method and Device for Allocating a Bit-budget Between Sub-frames in a CELP Codec," PCT patent application PCT / CA2018 / 51175

Claims

1. A system (100) for coding an object-based audio signal comprising audio objects (102) in response to audio streams with associated metadata, wherein the metadata of each audio object (102) comprise a number of metadata parameters and the audio objects (102) are processed by successive frames, the system (100) comprising: an audio stream processor (103) for analyzing the audio streams; a metadata processor (105) responsive to information on the audio streams from the analysis by the audio stream processor (103) for coding the metadata, wherein the metadata processor uses a logic for controlling a metadata coding bit-budget; and an encoder (109) for coding the audio streams; characterized in that: - the metadata processor (105) applies an inter-object metadata coding logic to use of absolute coding of the metadata parameters of different audio objects (102) to minimize, in a current frame, a number of metadata parameters of the different audio objects coded using absolute coding; and - the metadata processor (105), using the inter-object metadata coding logic, codes (a) a first metadata parameter of a first audio object using absolute coding in a frame M, (b) a second metadata parameter of the first audio object using absolute coding in a frame M+1, (c) the first metadata parameter of a second audio object using absolute coding in a frame M+2, and (d) the second metadata parameter of the second audio object using absolute coding in a frame M+3.

2. The system (100) according to claim 1, wherein the metadata processor (105) uses an intra-object metadata coding logic to limit absolute coding in a given frame to one metadata parameter, in order to limit a range of metadata coding bit-budget fluctuation between frames of the object-based audio signal and to avoid a lack of bit-budget left for coding the audio streams.

3. The system (100) according to claim 2, wherein the metadata processor (105), using the intra-object metadata coding logic, avoids in a same frame absolute coding of a first metadata parameter if a second metadata parameter was already coded using absolute coding.

4. The system (100) according to claim 2 or 3, wherein the intra-object metadata coding logic is bitrate dependent to enable absolute coding of a plurality of metadata parameters in the same frame.

5. The system (100) according to claim 1, wherein the metadata processor (105), using the inter-object metadata coding logic, controls frame counters of metadata parameters coded using absolute coding.

6. The system (100) according to claim 1 or 5, wherein the metadata processor (105), using the inter-object metadata coding logic, codes one audio object metadata parameter by frame.

7. The system (100) according to any one of claims 1, 5 and 6, wherein the inter-object metadata coding logic is bitrate dependent to enable absolute coding of a plurality of metadata parameters of the audio objects (102) in a same frame.

8. The system (100) according to any one of claims 1 to 7, wherein: - the audio stream processor (103) analyzes the audio streams to detect voice activity; - the metadata processor (105) comprises an analyzer of the metadata of each audio object (102) using the voice activity detection from the audio stream processor (103) to determine if a current frame is inactive or active with respect to the audio object (102); - in inactive frames, the metadata processor (105) codes no metadata relative to the audio object (102); and - in active frames, the metadata processor (105) codes the metadata for the audio object (102).

9. The system (100) according to any one of claims 1 to 8, wherein: the metadata of each audio object (102) comprise an azimuth parameter and an elevation parameter; and the metadata processor (105) comprises, to quantize the azimuth and elevation parameters, a quantizer of an azimuth index using an azimuth quantization step and of an elevation parameter index using an elevation quantization step.

10. The system (100) according to any one of claims 1 to 8, wherein: the metadata processor (105) comprises, to quantize a metadata parameter of an audio object (102), a quantizer of a metadata parameter index using a quantization step.

11. The system (100) according to any one of claims 1 to 10, wherein: the metadata of each audio object (102) comprise a plurality of metadata parameters; the metadata processor (105) represents the plurality of metadata parameters as one parameter; and the metadata processor (105) comprises a quantizer of an index of the said one parameter.

12. The system (100) according to any one of claims 1 to 8, wherein: the metadata processor (105) comprises, to quantize a metadata parameter of an audio object (102), a quantizer of a metadata parameter index using a quantization step; and the metadata processor (105) comprises a metadata encoder for coding the metadata parameter indexes using either absolute or differential coding.

13. The system (100) according to claim 12, wherein the metadata encoder codes the metadata parameter indexes using absolute coding if no metadata were present in a previous frame.

14. The system (100) according to claim 12 or 13, wherein the metadata encoder codes the metadata parameter indexes using absolute coding when a number of consecutive frames using differential coding is higher than a maximum number of consecutive frames coded using differential coding.

15. The system (100) according to any one of claims 12 to 14, wherein the metadata encoder, when coding a metadata parameter index using absolute coding, produces an absolute coding flag distinguishing between absolute and differential coding and followed by the metadata parameter index coded using absolute coding.

16. The system (100) according to claim 15, wherein the metadata encoder, when encoding a metadata parameter index using differential coding, sets the absolute coding flag to 0 and produces a zero coding flag following the absolute coding flag, signaling a difference between the metadata parameter index in a current frame and the metadata parameter index in a previous frame equal to 0.

17. The system (100) according to claim 16, wherein, if the difference between the metadata parameter index in the current frame and the metadata parameter index in the previous frame is not equal to 0, the metadata encoder produces a sign flag indicative of a plus or minus sign of the difference followed by a difference index indicative of the value of the difference.

18. The system (100) according to any one of claims 1 to 17, wherein the metadata processor (105) outputs information about bit-budgets for the coding of the metadata of the audio objects (102), and wherein the system (100) further comprises a bit-budget allocator (106) responsive to the information about the bit-budgets for the coding of the metadata of the audio objects (102) from the metadata processor (105) to allocate bitrates for the coding of the audio streams.

19. The system (100) according to claim 18, wherein the bit-budget allocator (106) sums the bit-budgets for the coding of the metadata of the audio objects (102), and adds the sum of the bit-budgets to a signaling bit-budget to perform bitrate distribution between the audio streams.

20. A method (150) for coding an object-based audio signal comprising audio objects (102) in response to audio streams with associated metadata, wherein the metadata of each audio object (102) comprise a number of metadata parameters and the audio objects (102) are processed by successive frames, the method (150) comprising: analyzing (153) the audio streams; coding (155) the metadata using information on the audio streams from the analysis (153) of the audio streams, and a logic for controlling a metadata coding bit-budget; and encoding (159) the audio streams; characterized in that: - using a logic for controlling the metadata coding bit-budget comprises applying an inter-object metadata coding logic to use of absolute coding of the metadata parameters of different audio objects (102) to minimize, in a current frame, a number of metadata parameters of the different audio objects (102) coded using absolute coding; and - using the inter-object metadata coding logic comprises coding (a) a first metadata parameter of a first audio object using absolute coding in a frame M, (b) a second metadata parameter of the first audio object using absolute coding in a frame M+1, (c) the first metadata parameter of a second audio object using absolute coding in a frame M+2, and (d) the second metadata parameter of the second audio object using absolute coding in a frame M+3.

21. The method (150) according to claim 20, wherein using a logic for controlling the metadata coding bit-budget comprises using an intra-object metadata coding logic to limit absolute coding in a given frame to one metadata parameter, in order to limit a range of metadata coding bit-budget fluctuation between frames of the object-based audio signal and to avoid a lack of bit-budget left for coding the audio streams.

22. The method (150) according to claim 21, wherein using the intra-object metadata coding logic comprises avoiding in a same frame absolute coding of a first metadata parameter if a second metadata parameter was already coded using absolute coding.

23. The method (150) according to any one of claims claim 21 or 22, wherein the intra-object metadata coding logic is bitrate dependent to enable absolute coding of a plurality of metadata parameters in the same frame.

24. The method (150) according to claim 20, wherein using the inter-object metadata coding logic comprises controlling frame counters of metadata parameters coded using absolute coding.

25. The method (150) according to claim, 20 or 24, wherein using the inter-object metadata coding logic comprises coding one audio object metadata parameter by frame.

26. The method (150) according to any one of claims 20, 24 and 25, wherein the inter-object metadata coding logic is bitrate dependent to enable absolute coding of a plurality of metadata parameters of the audio objects (102) in a same frame.

27. The method (150) according to any one of claims 20 to 26, comprising: - detecting voice activity upon analyzing the audio streams; - analyzing the metadata of each audio object (102) using the voice activity detection to determine if a current frame is inactive or active with respect to the audio object; - in inactive frames, encoding no metadata relative to the audio object; and - in active frames, encoding the metadata for the audio object.

28. The method (150) according to any one of claims 20 to 27, wherein: the metadata of each audio object (102) comprise an azimuth parameter and an elevation parameter; and quantizing the azimuth and elevation parameters comprises quantizing an azimuth index using an azimuth quantization step and quantizing an elevation parameter index using an elevation quantization step.

29. The method (150) according to any one of claims 20 to 27, comprising, to quantize a metadata parameter of an audio object (102), quantizing a metadata parameter index using a quantization step.

30. The method (150) according to any one of claims 20 to 29, wherein the metadata of each audio object comprise a plurality of metadata parameters, and wherein the method comprises: representing the plurality of metadata parameters as one parameter; and quantizing an index of the said one parameter.

31. The method (150) according to any one of claims 20 to 27, wherein: the method (150) comprises, to quantize a metadata parameter of an audio object (102), quantizing a metadata parameter index using a quantization step; the method (150) comprises coding the metadata parameter indexes using either absolute or differential coding: and coding the metadata parameter indexes comprises using absolute coding if no metadata were present in a previous frame.

32. The method (150) according to any one of claims 20 to 27, wherein: the method (150) comprises, to quantize a metadata parameter of an audio object (102), quantizing a metadata parameter index using a quantization step: the method (150) comprises coding the metadata parameter indexes using either absolute or differential coding: and coding the metadata parameter indexes comprises using absolute coding when a number of consecutive frames using differential coding is higher than a maximum number of consecutive frames coded using differential coding.

33. The method (150) according to any one of claims 20 to 27, wherein: the method (150) comprises, to quantize a metadata parameter of an audio object, quantizing a metadata parameter index using a quantization step; the method (150) comprises coding the metadata parameter indexes using either absolute or differential coding; and coding a metadata parameter index using absolute coding comprises producing an absolute coding flag distinguishing between absolute and differential coding and followed by the metadata parameter index coded using absolute coding.

34. The method (150) according to claim 33, wherein coding a metadata parameter index using differential coding comprises setting the absolute coding flag to 0 and producing a zero coding flag following the absolute coding flag, signaling a difference between the metadata parameter index in a current frame and the metadata parameter index in a previous frame equal to 0.

35. The method (150) according to claim 34, wherein coding a metadata parameter index using differential coding comprises, if the difference between the metadata parameter index in the current frame and the metadata parameter index in the previous frame is not equal to 0, producing a sign flag indicative of a plus or minus sign of the difference followed by a difference index indicative of the value of the difference.

36. The method (150) according to any one of claims 20 to 35, wherein coding (155) the metadata comprises outputting information about bit-budgets for the coding of the metadata of the audio objects (102), and wherein the method (150) comprises a bit-budget allocation (156) responsive to the information about the bit-budgets for the coding of the metadata of the audio objects (102) to allocate bitrates for the coding of the audio streams.

37. The method (150) according to claim 36, wherein the bit-budget allocation (156) comprises summing the bit-budgets for the coding of the metadata of the audio objects (102), and adding the sum of the bit-budgets to a signaling bit-budget to perform bitrate distribution between the audio streams.