INFORMATION REQUIRED FOR CODING A HIGHER ORDER AMBISONICS REPRESENTATION OF A SOUND FIELD Technical field

The invention relates to a method and to an apparatus for improving the coding of side information required for coding a Higher Order Ambisonics representation of a sound field.

Background

Higher Order Ambisonics (HOA) offers one possibility to rep- resent three-dimensional sound among other techniques like wave field synthesis (WFS) or channel based approaches like the 22.2 multichannel audio format. In contrast to channel based methods, the HOA representation offers the advantage of being independent of a specific loudspeaker set-up. This flexibility, however, is at the expense of a decoding pro ^¬ cess which is required for the playback of the HOA representation on a particular loudspeaker set-up. Compared to the WFS approach, where the number of required loudspeakers is usually very large, HOA signals may also be rendered to set- ups consisting of only few loudspeakers. A further advantage of HOA is that the same representation can also be employed without any modification for binaural rendering to headphones .

HOA is based on the representation of the spatial density of complex harmonic plane wave amplitudes by a truncated Spher ^¬ ical Harmonics (SH) expansion. Each expansion coefficient is a function of angular frequency, which can be equivalently represented by a time domain function. Hence, without loss of generality, the complete HOA sound field representation actually can be assumed to consist of 0 time domain func ^¬ tions, where 0 denotes the number of expansion coefficients. These time domain functions will be equivalently referred to as HOA coefficient sequences or as HOA channels in the fol- lowing.

The spatial resolution of the HOA representation improves with a growing maximum order N of the expansion. Unfortunately, the number of expansion coefficients 0 grows quad- ratically with the order N, in particular 0 = (N + l) ² . For example, typical HOA representations using order N = 4 re ^¬ quire 0 = 25 HOA (expansion) coefficients. According to the previously made considerations, the total bit rate for the transmission of HOA representation, given a desired single- channel sampling rate f _$ and the number of bits per sam- pie, is determined by 0 · f _s · . Consequently, transmitting an HOA representation of order N = 4 with a sampling rate of fs = 48kHz employing = 16 bits per sample results in a bit rate of 19.2MBits/s, which is very high for many practical applications like e.g. streaming. Thus, compression of HOA representations is highly desirable.

The compression of HOA sound field representations is proposed in WO 2013/171083 Al, EP 13305558.2 and PCT/EP2013/075559. These processings have in common that they perform a sound field analysis and decompose the given HOA representation into a directional component and a residual ambient compo ^¬ nent. On one hand the final compressed representation is as ^¬ sumed to consist of a number of quantised signals, resulting from the perceptual coding of the directional signals and relevant coefficient sequences of the ambient HOA component. On the other hand it is assumed to comprise additional side information related to the quantised signals, which side in ^¬ formation is necessary for the reconstruction of the HOA representation from its compressed version. An important part of that side information is a description of a prediction of portions of the original HOA representa ^¬ tion from the directional signals. Since for this prediction the original HOA representation is assumed to be equivalent- ly represented by a number of spatially dispersed general plane waves impinging from spatially uniformly distributed directions, the prediction is referred to as spatial predic ^¬ tion in the following.

The coding of such side information related to spatial pre- diction is described in ISO/IEC JTC1 /SC29/WG11 , N14061,

"Working Draft Text of MPEG-H 3D Audio HOA RMO", November 2013 , Geneva, Switzerland. However, this state-of-the-art coding of the side information is rather inefficient.

Summary of invention

A problem to be solved by the invention is to provide a more efficient way of coding side information related to that spatial prediction.

This problem is solved by the methods disclosed in claims 1 and 6. An apparatus that utilises these methods is disclosed in claims 2 and 7. A bit is prepended to the coded side information representa ^¬ tion data ζcoΌr which bit signals whether or not any predic ^¬ tion is to be performed. This feature reduces over time the average bit rate for the transmission of the ^COD data. Fur ^¬ ther, in specific situations, instead of using a bit array indicating for each direction if the prediction is performed or not, it is more efficient to transmit or transfer the number of active predictions and the respective indices. A single bit can be used for indicating in which way the indi ^¬ ces of directions are coded for which a prediction is sup- posed to be performed. On average, this operation over time further reduces the bit rate for the transmission of the ζοο _ϋ data . In principle, the inventive method is suited for improving the coding of side information required for coding a Higher Order Ambisonics representation of a sound field, denoted HOA, with input time frames of HOA coefficient sequences, wherein dominant directional signals as well as a residual ambient HOA component are determined and a prediction is used for said dominant directional signals, thereby provid ^¬ ing, for a coded frame of HOA coefficients, side information data describing said prediction, and wherein said side information data can include:

- a bit array indicating whether or not for a direction a prediction is performed;

a bit array in which each bit indicates, for the direc ^¬ tions where a prediction is to be performed, the kind of the prediction;

- a data array whose elements denote, for the predictions to be performed, indices of the directional signals to be used;

a data array whose elements represent quantised scaling factors ,

said method including the step:

providing a bit value indicating whether or not said prediction is to be performed;

if no prediction is to be performed, omitting said bit arrays and said data arrays in said side information data; - if said prediction is to be performed, providing a bit value indicating whether or not, instead of said bit array indicating whether or not for a direction a prediction is performed, a number of active predictions and a data array containing the indices of directions where a prediction is to be performed are included in said side information data

In principle the inventive apparatus is suited for improving the coding of side information required for coding a Higher Order Ambisonics representation of a sound field, denoted HOA, with input time frames of HOA coefficient sequences, wherein dominant directional signals as well as a residual ambient HOA component are determined and a prediction is used for said dominant directional signals, thereby provid- ing, for a coded frame of HOA coefficients, side information data describing said prediction, and wherein said side information data can include:

a bit array indicating whether or not for a direction a prediction is performed;

- a bit array in which each bit indicates, for the direc ^¬ tions where a prediction is to be performed, the kind of the prediction;

a data array whose elements denote, for the predictions to be performed, indices of the directional signals to be used;

a data array whose elements represent quantised scaling factors ,

said apparatus including means which:

provide a bit value indicating whether or not said pre- diction is to be performed;

if no prediction is to be performed, omit said bit arrays and said data arrays in said side information data;

if said prediction is to be performed, provide a bit val ^¬ ue indicating whether or not, instead of said bit array in- dicating whether or not for a direction a prediction is performed, a number of active predictions and a data array con ^¬ taining the indices of directions where a prediction is to be performed are included in said side information data. Advantageous additional embodiments of the invention are disclosed in the respective dependent claims.

Brief description of drawings

Exemplary embodiments of the invention are described with reference to the accompanying drawings, which show in:

Fig. 1 Exemplary coding of side information related to spa- tial prediction in the HOA compression processing described in EP 13305558.2;

Fig. 2 Exemplary decoding of side information related to spatial prediction in the HOA decompression processing described in patent application EP 13305558.2; Fig. 3 HOA decomposition as described in patent application

PCT/EP2013/075559;

Fig. 4 Illustration of directions (depicted as crosses) of general plane waves representing the residual signal and the directions (depicted as circles) of dominant sound sources. The directions are presented in a three-dimensional coordinate system as sampling po ^¬ sitions on the unit sphere;

Fig. 5 State of art coding of spatial prediction side in ^¬ formation;

Fig. 6 Inventive coding of spatial prediction side information;

Fig. 7 Inventive decoding of coded spatial prediction side information;

Fig. 8 Continuation of Fig. 7.

Description of embodiments

In the following, the HOA compression and decompression pro- cessing described in patent application EP 13305558.2 is recapitulated in order to provide the context in which the in ^¬ ventive coding of side information related to spatial pre ^¬ diction is used.

HOA compression

In Fig. 1 it is illustrated how the coding of side information related to spatial prediction can be embedded into the HOA compression processing described patent application EP 13305558.2. For the HOA representation compression, a frame-wise processing with non-overlapping input frames C(/c) of HOA coeffi ^¬ cient sequences of length L is assumed, where k denotes the frame index. The first step or stage 11/12 in Fig. 1 is op ^¬ tional and consists of concatenating the non-overlapping k- th and ( k— 1) -th frames of HOA coefficient sequences C(/c) in ^¬ to a long frame C(/c) as

C(fc):= [C(fc-l) C{k)} , (1) which long frame is 50% overlapped with an adjacent long frame and which long frame is successively used for the es ^¬ timation of dominant sound source directions. Similar to the notation for C(/c), the tilde symbol is used in the following description for indicating that the respective quantity re ^¬ fers to long overlapping frames. If step/stage 11/12 is not present, the tilde symbol has no specific meaning.

A parameter in bold means a set of values, e.g. a matrix or a vector.

The long frame C(/c) is successively used in step or stage 13 for the estimation of dominant sound source directions as described in EP 13305558.2. This estimation provides a data set JDIR _, ACT(k) <Ξ {1, ... , D} of indices of the related directional signals that have been detected, as well as a data set

5^ _ACT (/c) of the corresponding direction estimates of the directional signals. D denotes the maximum number of direc- tional signals that has to be set before starting the HOA compression and that can be handled in the known processing which follows.

In step or stage 14, the current (long) frame C(/c) of HOA co- efficient sequences is decomposed (as proposed in EP 13305156.5) into a number of directional signals X _mR (k— 2) belonging to the directions contained in the set Qa,Acr(k , and a residual ambient HOA component C _AMB (k— 2). The delay of two frames is introduced as a result of overlap-add processing in order to obtain smooth signals. It is assumed that X _mR (k— 2) is con ^¬ taining a total of D channels, of which however only those corresponding to the active directional signals are non ^¬ zero. The indices specifying these channels are assumed to be output in the data set _mRACT (k— 2). Additionally, the de- composition in step/stage 14 provides some parameters (/c— 2) which can be used at decompression side for predicting portions of the original HOA representation from the directional signals (see EP 13305156.5 for more details) . In order to explain the meaning of the spatial prediction parameters ζίΗ— 2), the HOA decomposition is described in more detail in the below section HOA decomposition .

In step or stage 15, the number of coefficients of the ambi ^¬ ent HOA component C _AMB (k— 2) is reduced to contain only

ORED + D— N _mRACT (k— 2) non-zero HOA coefficient sequences, where N _mRACT (k— ^— 2)| indicates the cardinality of the data set _mRACT (k— 2), i.e. the number of active di ^¬ rectional signals in frame k— 2. Since the ambient HOA com ^¬ ponent is assumed to be always represented by a minimum num ^¬ ber ORED of HOA coefficient sequences, this problem can be actually reduced to the selection of the remaining D— N _mRACT (k— 2) HOA coefficient sequences out of the possible 0— 0 _REO ones. In order to obtain a smooth reduced ambient HOA representa ^¬ tion, this choice is accomplished such that, compared to the choice taken at the previous frame k— 3, as few changes as possible will occur.

The final ambient HOA representation with the reduced number of 0 _RED + N _DIRiACT (/c— 2) non-zero coefficient sequences is de- noted by C _{AMB REO} (k— 2) . The indices of the chosen ambient HOA coefficient sequences are output in the data set 2) . In step/stage 16, the active directional signals contained in X _mR (k— 2) and the HOA coefficient sequences contained in C _AMBjRED (/c— 2) are assigned to the frame Y(k— 2) of / channels for individual perceptual encoding as described in EP 13305558.2. Perceptual coding step/stage 17 encodes the / channels of frame Y(k— 2) and outputs an encoded frame Y(k— 2) .

According to the invention, following the decomposition of the original HOA representation in step/stage 14, the spa ^¬ tial prediction parameters or side information data (/c— 2) resulting from the decomposition of the HOA representation are losslessly coded in step or stage 19 in order to provide a coded data representation ^ 2), using the index set

^) delayed by two frames in delay 18.

HOA decompression

In Fig. 2 it is exemplary shown how to embed in step or stage 25 the decoding of the received encoded side infor- mation data ^ 2) related to spatial prediction into the HOA decompression processing described in Fig. 3 of patent application EP 13305558.2. The decoding of the encoded side information data ^ 2) is carried out before entering its decoded version (/c— 2) into the composition of the HOA representation in step or stage 23, using the received index set _mRACT (k) delayed by two frames in delay 24.

In step or stage 21 a perceptual decoding of the / signals contained in Y(k— 2) is performed in order to obtain the / decoded signals in Y(k— 2) .

In signal re-distributing step or stage 22, the perceptually decoded signals in Y(k— 2) are re-distributed in order to recreate the frame X _mR (k— 2) of directional signals and the frame C _{AMB RED} (k— 2) of the ambient HOA component. The infor ^¬ mation about how to re-distribute the signals is obtained by reproducing the assigning operation performed for the HOA compression, using the index data sets _{mR ACT} (k) ^an d ^AMB.ACT C^ ^— 2) . In composition step or stage 23, a current frame C(k— 3) of the desired total HOA representation is re-composed (accord ^¬ ing to the processing described in connection with Fig. 2b and Fig. 4 of PCT/EP2013/075559 using the frame X _mR {k - 2) of the directional signals, the set ^DIRACT C^) °f the active di ^¬ rectional signal indices together with the set (fc) of the corresponding directions, the parameters (/c— 2) for pre ^¬ dicting portions of the HOA representation from the directional signals, and the frame C _AMBREO (k— 2) of HOA coefficient sequences of the reduced ambient HOA component.

C _AMBJRED (/c— 2) corresponds to component D _A (k— 2) in PCT/EP2013/ 075559, and Gn,A j (k ^{A N CL} ^DIRACT C^) correspond to A^ (k) in PCT/ EP2013/075559, wherein active directional signal indices can be obtained by taking those indices of rows of A^ k which contain valid elements. I.e., directional signals with re ^¬ spect to uniformly distributed directions are predicted from the directional signals X _mR (k— 2) using the received parame ^¬ ters (/c— 2) for such prediction, and thereafter the current decompressed frame C(k— 3) is re-composed from the frame of directional signals X _mR (k— 2) , from _{mR ACT} (k) and OO r and from the predicted portions and the reduced ambient HOA compo- nent CAMB.RED (fc - 2) . HOA decomposition

In connection with Fig. 3 the HOA decomposition processing is described in detail in order to explain the meaning of the spatial prediction therein. This processing is derived from the processing described in connection with Fig. 3 of patent application PCT/EP2013/075559.

First, the smoothed dominant directional signals X _mR (k— 1) and their HOA representation C _DIR (/c— 1) are computed in step or stage 31, using the long frame C(/c) of the input HOA rep- resentation, the set of directions and the set ¾IR,ACT of corresponding indices of directional signals. It is as ^¬ sumed that X _mR (k— 1) contains a total of D channels, of which however only those corresponding to the active directional signals are non-zero. The indices specifying these channels are assumed to be output in the set _{mR ACT} (k— 1) . In step or stage 33 the residual between the original HOA representation C(k— 1) and the HOA representation C _DIR (/c— 1) of the dominant directional signals is represented by a num ^¬ ber of 0 directional signals -X _RES (/C— 1), which can be consid- ered as being general plane waves from uniformly distributed directions, which are referred to a uniform grid.

In step or stage 34 these directional signals are predicted from the dominant directional signals X _mR (k— 1) in order to provide the predicted signals -X _RES (/C— 1) together with the respective prediction parameters (/c— 1) . For the prediction only the dominant directional signals Xum _,d (k— 1) with indices d , which are contained in the set ^DIRACT C^ ^— Or ^are consid ^¬ ered. The prediction is described in more detail in the be ^¬ low section Spatial prediction.

In step or stage 35 the smoothed HOA representation C _RES (/c— 2) of the predicted directional signals X _RE$ (k— 1) is computed. In step or stage 37 the residual C _AMB (/c— 2) between the orig- inal HOA representation C(k— 2) and the HOA representation C _DIR (/c— 2) of the dominant directional signals together with the HOA representation C _RES (/c— 2) of the predicted directional signals from uniformly distributed directions is computed and is output.

The required signal delays in the Fig. 3 processing are per ^¬ formed by corresponding delays 381 to 387.

Spatial prediction

The goal of the spatial prediction is to predict the 0 re ^¬ sidual signals

from the extended frame

X _mR (k - 1): = [X _OlR (k - 3) X _OlR (k - 2) X _OlR (k - 1)] (3)

of smoothed directional signals (see the description in above section HOA decomposition and in patent application PCT/EP2013/075559) .

Each residual signal ¾ES,GRiD _,q ^— 1) r q ⁼ t,—,0, represents a spatially dispersed general plane wave impinging from the direction _q , whereby it is assumed that all the directions q, q = l,...,0 are nearly uniformly distributed over the unit sphere. The total of all directions is referred to as a ' grid ' .

Each directional signal ¾IR _,C J ^~ 1) _> d = 1,...,D represents a general plane wave impinging from a trajectory interpolated between the directions Ω _ΑσΓιά & - 3), Q _ACTid {k - 2) _r n _ACTid (k - 1) and Ω _{Α Τ d} (k) , assuming that the d-th directional signal is active for the respective frames.

To illustrate the meaning of the spatial prediction by means of an example, the decomposition of an HOA representation of order N = 3 is considered, where the maximum number of direc ^¬ tions to extract is equal to D =4. For simplicity it is fur ^¬ ther assumed that only the directional signals with indices '1' and '4' are active, while those with indices '2' and '3' are non-active. Additionally, for simplicity it is assumed that the directions of the dominant sound sources are constant for the considered frames, i.e. ii _ACTd (k— 3) =

12 _ACT , _d k) = 12 _ACT , _d for d = 1,4 ( 5 )

As a consequence of order N = 3, there are 0 = 16 directions 12 _q of spatially dispersed general plane waves ¾ _ES ,GRiD _,q ^— 1) / q = l,...,0. Fig. 4 shows these directions together with the directions 12 _ACT ,ι ^an d 12 _ACT ,4 °f the active dominant sound sources.

State-of-the-art parameters for describing the spatial pre ^¬ diction

One way of describing the spatial prediction is presented in the above-mentioned ISO/IEC document. In this document, the signals ¾ _ES ,GRiD _,q ^— 1) / q = l,...,0 are assumed to be predicted by a weighted sum of a predefined maximum number D _{P ED} °f di ^¬ rectional signals, or by a low pass filtered version of the weighted sum. The side information related to spatial pre ^¬ diction is described by the parameter set ^(/c— 1) = {p _TYPE (/c— l),P _IND (/c— l),Pq _F (k— 1)}, which consists of the following three components :

• The vector p _TYPE (/c— 1) whose elements _TYPE, ^ ^— 1) _/ q = l,...,0 indicate whether or not for the q- direction _q a pre ^¬ diction is performed, and if so, then they also indicate which kind of prediction. The meaning of the elements is as follows: (0 for no prediction for direction _q

1 for a full band prediction for direction H _q . ( 6 )

2 for a low band prediction for direction H _q

• The matrix P _IND (/c— 1), whose elements Pi _ND A _q ^~ 1)'

d = 1, ... , Dp ED t 9 = 1, denote the indices from which di ^¬ rectional signals the prediction for the direction _q has to be performed. If no prediction is to be performed for a direction _q , the corresponding column of the matrix ^ _IND C^- 1) consists of zeros. Further, if less than D _PRED directional signals are used for the prediction for a di ^¬ rection _q , the non-required elements in the 9-th column of P _IND (/c— 1) are also zero.

• The matrix P _QF (/C— 1), which contains the corresponding

quantised prediction factors P _Q,F , _d , _q ^— 1) r d = 1, ... , D _PRED r q = 1, ...,0.

The following two parameters have to be known at decoding side for enabling the appropriate interpretation of these parameters :

• The maximum number D _{P ED} °f directional signals, from

which a general plane wave signal ¾ _ES ,GRiD, _q ~ 1) is allowed to be predicted.

· The number B _sc of bits used for quantising the prediction factors _Q,F , _d , _q ^— 1) r d = 1, ... , Dp ED r 9 = 1, ...,0. The de-quanti ^¬ sation rule is given in equation (10) .

These two parameters have to either be set to fixed values known to the encoder and decoder, or to be additionally transmitted, but distinctly less frequently than the frame rate. The latter option may be used for adapting the two pa ^¬ rameters to the HOA representation to be compressed.

An example for a parameter set may look like the following, assuming 0 = 16, D _{P ED} ⁼ 2 and B _sc = 8:

P _TYPE O -1) = [1 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0] , (7) r _j , _ _Λ Λ - Γ ¹ 0 0 0 0 0 1 0 0 0 0 0 0 0 0 01

^IND ^ ^} Lo 0 0 0 0 0 4 0 0 0 0 0 0 0 0 oJ ' ^{( )} p _(k _ ₌ [40 0 0 0 0 0 15 0 0 0 0 0 0 0 0 01

Q' ⁾ I- 0 0 0 0 0 0 - 13 0 0 0 0 0 0 0 0 0-1 ^{' ( )}

Such parameters would mean that the general plane wave sig ^¬ nal ¾ES,GRID,I ^— 1) from direction Ω is predicted from the di- rectional signal ¾IR,I(^ ^— Ό from direction -QACT.I by a pure multiplication (i.e. full band) with a factor that results from de-quantising the value 40. Further, the general plane wave signal ¾ES,GRID,7 ^— 1) from direction Ω ₇ is predicted from the directional signals x _DIR1 (/c— 1) and ¾iR,4(k ^— 1) by a lowpass filtering and multiplication with factors that result from de-quantising the values 15 and —13.

Given this side information, the prediction is assumed to be performed as follows:

First, the quantised prediction factors _Q ,F _,d,q ^— 1) r

d = 1, ... , Dp ED _r q = l,...,0 are dequantised to provide the actual prediction factors

Ρ,_,(*-Ι) - ^{1) +} ») ^2~ "~ ^{+1 lf} » *-i>*o _ ₍₁₀₎ i ^f lNDAqO - 1) = 0

As already mentioned, B _sc denotes a predefined number of bits to be used for the quantisation of the prediction factors. Additionally, PF _,d, q.k— 1) is assumed to be set to zero, if iND,d,q(k ^— 1) is equal to zero.

For the previously mentioned example, assuming B _sc = 8, the de-quantised prediction factor vector would result in

,, _ _ι Γ0.3164 0 0 0 0 0 0.1211 0 0 0 0 0 0 0 0 0]

} ^¾ L 0 0 0 0 0 0 - 0.0977 0 0 0 0 0 0 0 0 0-1 ^{' ( )}

Further, for performing a low pass prediction a predefined low pass FIR filter ft _LP : = [¾ _LP (0) ¾ _LP (1) ... h _LP (L _h - 1)] (12) of length L _h = 31 is used. The filter delay is given by D _h = 15 samples .

Assuming as signals the predicted signals and the direction

to be composed of their samples by ¾ES,q (k ^— 1) =

[¾ES,q (k - 1,1) ¾ES,q (k - 1.2) ... (fc - 1,2L)] for q = 1, ... , 0 , (15)

[¾iR. _£ t(fr - LI) ¾iR. _£ i(fc - 1»2) ... x _DIRid (fc - 1,3L)] for d = 1, ...,D , (16) the sample values of the predicted signals are given by

(17)

with y _LPi(? (fc - 1,0 :=

_∑ min(L _h -l,l ₊ 2D _h -l) . ^^^(fc - 1, L + Z + D _h - j . (18)

As already mentioned and as now can be seen from equation (17), the signals ¾RES,GRiD,q ^— 1)/ q = l, ... , 0 are assumed to be predicted by a weighted sum of a predefined maximum number ^PRED °f directional signals, or by a low pass filtered ver ^¬ sions of the weighted sum.

State-of-the-art coding of the side information related to spatial prediction

In the above-mentioned ISO/IEC document the coding of the spatial prediction side information is addressed. It is sum' marised in Algorithm 1 depicted in Fig. 5 and will be ex ^¬ plained in the following. For a clearer presentation the frame index k— 1 is neglected in all expressions.

First, a bit array ActivePred consisting of 0 bits is creat- ed, in which the bit ActivePred [q>] indicates whether or not for the direction _q a prediction is performed. The number of 'ones' in this array is denoted by NumActivePred .

Next, the bit array PredType of length NumActivePred is creat- ed where each bit indicates, for the directions where a pre ^¬ diction is to be performed, the kind of the prediction, i.e. full band or low pass. At the same time, the unsigned inte ^¬ ger array PredDirSiglds of length NumActivePred · D _PRED is created, whose elements denote for each active prediction the

OpRED indices of the directional signals to be used. If less than DpRED directional signals are to be used for the predic ^¬ tion, the indices are assumed to be set to zero. Each ele ^¬ ment of the array PredDirSiglds is assumed to be represented by [log ₂ (D + 1)1 bits. The number of non-zero elements in the array PredDirSiglds is denoted by NumNonZerolds .

Finally, the integer array QuantPredGains of length NumNonZerolds is created, whose elements are assumed to represent the quantised scaling factors f _Q ,F _,d,q (k ^— 1) to be used in equation (17) . The dequantisation to obtain the corresponding dequan- tised scaling factors P _Fdq (k— l) is given in equation (10). Each element of the array QuantPredGains is assumed to be represented by B _sc bits.

In the end, the coded representation of the side information ^COD consists of the four aforementioned arrays according to ζ _£0Ό = [ActivePred PredType PredDirSiglds QuantPredGains] . (19) For explaining this coding by an example, the coded repre ^¬ sentation of equations (7) to (9) is used:

ActivePred = [1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0] (20) PredType = [0 1] (21) PredDirSiglds = [1 0 1 4] (22)

QuantPredGains = [40 15 -13] . (23) The number of required bits is equal to 16 + 2 + 3 · 4 + 8 · 3 = 54. Inventive coding of the side information related to spatial prediction

In order to increase the efficiency of the coding of the side information related to spatial prediction, the state- of-the-art processing is advantageously modified.

A) When coding HOA representations of typical sound scenes, the inventors have observed that there are often frames where in the HOA compression processing the decision is taken to not perform any spatial prediction at all. How- ever, in such frames the bit array ActivePred consists of zeros only, the number of which is equal to 0. Since such frame content occurs quite often, the inventive pro ^¬ cessing prepends to the coded representation ^COD ^a single bit PSPredictionActive, which indicates if any prediction is to be performed or not. If the value of the bit

PSPredictionActive is zero (or '1' as an alternative), the array ActivePred and further data related to the prediction are not to be included into the coded side information ^COD · I ^R practise, this operation reduces over time the average bit rate for the transmission of ^COD ·

B) A further observation made while coding HOA representations of typical sound scenes is that the number

NumActivePred of active prediction is often very low. In such situation, instead of using the bit array ActivePred for indicating for each direction _q whether or not the prediction is performed, it can be more efficient to transmit or transfer instead the number of active predic ^¬ tions and the respective indices. In particular, this modified kind of coding the activity is more efficient in case that NumActivePred < M _M , (24) where M _M is the greatest integer number that satisfies

[log ₂ (M _M )1 + M _M · [log ₂ (0)1 < 0 . (25) The value of M _M can be computed only with the knowledge of the HOA order N : 0 = (N + l) ² as mentioned above.

In equation (25) , [log ₂ (M _M )l denotes the number of bits re ^¬ quired for coding the actual number NumActivePred of active predictions, and M _M · [log ₂ (0)l is the number of bits re ^¬ quired for coding the respective direction indices. The right hand side of equation (25) corresponds to the num ^¬ ber of bits of the array ActivePred , which would be re ^¬ quired for coding the same information in the known way. According to the aforementioned explanations, a single bit KindOfCodedPredlds can be used for indicating in which way the indices of those directions, where a prediction is supposed to be performed, are coded. If the bit

KindOfCodedPredlds has the value '1' (or '0' in the alterna- tive) , the number NumActivePred and the array Predlds containing the indices of directions, where a prediction is supposed to be performed, are added to the coded side in ^¬ formation ^COD · Otherwise, if the bit KindOfCodedPredlds has the value '0' (or '1' in the alternative), the array

ActivePred is used to code the same information.

On average, this operation reduces over time the bit rate for the transmission of ^COD · To further increase the side information coding efficien- cy, the fact is exploited that often the actually availa ^¬ ble number of active directional signals to be used for prediction is less than D. This means that for the coding of each element of the index array PredDirSiglds less than [log ₂ (D + 1)1 bits are required. In particular, the actually available number of active directional signals to be used for prediction is given by the number D _ACT of elements of the data set _mRACT , which contains the indices IACT.I'■■■ ' CT,D _ACT of the active directional signals. Hence, log ₂ (|OACT + l| bits can be used for coding each element of the index ar ^¬ ray PredDirSiglds , which kind of coding is more efficient. In the decoder the data set ¾IR,ACT is assumed to be known, and thus the decoder also knows how many bits have to be read for decoding an index of a directional signal. Note that the frame indices of ^COD to be computed and the used index data set ¾IR,ACT have to be identical.

The above modifications A) to C) for the known side infor ^¬ mation coding processing result in the example coding processing depicted in Fig. 6.

Consequently, the coded side information consists of the following components: ^COD ⁼ (26)

/[PSPredictionActive] if PSPredictionActive = 0

' "PSPredictionActive

KindOfCodedPredlds

ActivePred

if PSPredictionActive = 1 Λ KindOfCodedPredlds = 0

PredType

PredDirSiglds

QuantPredGains

PSPredictionActive

KindOfCodedPredlds

NumActivePred

Predlds if PSPredictionActive = 1 Λ KindOfCodedPredlds

PredType

PredDirSiglds

QuantPredGains

Remark: in the above-mentioned ISO/IEC document e.g. in sec ^¬ tion 6.1.3, QuantPredGains is called PredGains , which however contains quantised values.

The coded representation for the example in equations (7) to (9) would be:

PSPredictionActive = 1 (27)

KindOfCodedPredlds = 1 (28)

NumActivePred = 2 (29)

Predlds = [1 7] (30) PredType = [0 1] ( 31 ) PredDirSiglds = [1 0 1 4] ( 32 ) QuantPredGains = [40 15 -13] , ( 33 ) and the required number of bits is 1 + 1 + 2 + 2 · 4 + 2 + 2 · 4 + 8 · 3 = 46 . Advantageously, compared to the state of the art coded rep ^¬ resentation in equations ( 2 0 ) to ( 23 ) , this representation coded according to the invention requires 8 bits less.

It is also possible to not provide bit array PredType at en ^¬ coder side.

Decoding of the modified side information coding related to spatial prediction

The decoding of the modified side information related to spatial prediction is summarised in the example decoding processing depicted in Fig. 7 and Fig. 8 (the processing depicted in Fig. 8 is the continuation of the processing depicted in Fig. 7 ) and is explained in the following.

Initially, all elements of vector p _TYPE and matrices J^ND ^an d PQ _F are initialised by zero. Then the bit PSPredictionActive is read, which indicates if a spatial prediction is to be per ^¬ formed at all. In the case of a spatial prediction (i.e.

PSPredictionActive = 1 ) , the bit KindOfCodedPredlds is read, which indicates the kind of coding of the indices of directions for which a prediction is to be performed.

In the case that KindOfCodedPredlds = 0 , the bit array ActivePred of length 0 is read, of which the q-t element indicates if for the direction _Q a prediction is performed or not. In a next step, from the array ActivePred the number NumActivePred of predictions is computed and the bit array PredType of length NumActivePred is read, of which the elements indicate the kind of prediction to be performed for each of the rele ^¬ vant directions. With the information contained in ActivePred and PredType , the elements of the vector TYPE ^are computed. It is also possible to not provide bit array PredType at en ^¬ coder side and to compute the elements of vector PTYPE from bit array ActivePred .

In case KindOfCodedPredlds = 1 , the number NumActivePred of active predictions is read, which is assumed to be coded with [log ₂ (M _M )l bits, where M _M is the greatest integer number satisfying equation (25) . Then, the data array Predlds consisting of

NumActivePred elements is read, where each element is assumed to be coded by [log ₂ (0)l bits. The elements of this array are the indices of directions, where a prediction has to be per ^¬ formed. Successively, the bit array PredType of length

NumActivePred is read, of which the elements indicate the kind of prediction to be performed for each one of the relevant directions. With the knowledge of NumActivePred, Predlds and PredType, the elements of the vector PTYPE ^are computed.

It is also possible to not provide bit array PredType at en ^¬ coder side and to compute the elements of vector TYPE from number NumActivePred and from data array Predlds .

For both cases (i.e. KindOfCodedPredlds = 0 and KindOfCodedPredlds = 1 ) , in the next step the array PredDirSiglds is read, which con ^¬ sists of NumActivePred · D _PRED elements. Each element is assumed to be coded by log ₂ (OACT)l bits. Using the information con ^¬ tained in PTYPE ¾IR _, ACT ^an d PredDirSiglds , the elements of ma ^¬ trix PJ D are set and the number NumNonZeroIds of non-zero el- ements in P _IND is computed.

Finally, the array QuantPredGains is read, which consists of NumNonZeroIds elements, each coded by B _SC bits. Using the information contained in and QuantPredGains , the elements of the matrix P _QF are set.

The inventive processing can be carried out by a single pro ^¬ cessor or electronic circuit, or by several processors or electronic circuits operating in parallel and/or operating on different parts of the inventive processing.