Apparatus and method for encoding or decoding multichannel signals using side gain and residual gain

Abstract

To provide an improved concept for processing multi-channel data.SOLUTION: A device for encoding a multi-channel signal 100 including at least two channels 101, 102, includes: a downmixer 120 for calculating a down-mix signal 122 from the multi-channel signal 100; a parameter calculator 140 for calculating a side gain 141 from a first channel 101 of the at least two channels and a second channel 102 of the at least two channels, and calculating a residual gain 142 from the first channel 101 and the second channel 102; and an output interface 160 for generating an output signal, the output signal containing information concerning the down-mix signal 122 and information concerning the side gain 141 and the residual gain 142.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to the field of audio encoding, and in particular, to the field of stereo or multi-channel encoding / decoding. Background Art

[0002] State-of-the-art methods for lossy parametric encoding of stereo signals at low bit rates are based on parametric stereo as standardized in MPEG-4 Part 3. The general idea is to reduce the number of channels by computing a downmix signal from two input channels after extracting stereo parameters that are transmitted to a decoder as side information. These stereo parameters are typically inter-channel level difference (ILD), inter-channel phase difference (IPD), and inter-channel coherence (ICC), which are calculated in subbands and capture the spatial image to a certain extent.

[0003] The decoder performs upmixing of the mono input to create two channels satisfying the relationships of ILD, IPD, and ICC. This is done by matrixing the input signal together with a decorrelated version of the signal generated at the decoder.

[0004] For example, it has been found that the use of such parameters introduces considerable complexity for calculating and processing these parameters. Specifically, the ILD parameter is problematic because it can have very small or very large values, and this almost unlimited range of values causes problems for efficient calculation, quantization, and the like. Summary of the Invention Problem to be Solved by the Invention

[0005] An object of the present invention is to provide an improved concept for processing multi-channel data. Means for Solving the Problem

[0006] This objective is achieved by an apparatus for encoding a multichannel signal according to claim 1, an apparatus for decoding an encoded multichannel signal according to claim 25, a method for encoding a multichannel signal according to claim 41, a method for decoding an encoded multichannel signal according to claim 42, a computer program according to claim 43, or an encoded multichannel signal according to claim 44.

[0007] In a first aspect of the present invention, in contrast to the prior art, a different parametric coding procedure is employed that depends on two gain parameters, namely a side gain parameter and a residual gain parameter. Both gain parameters are calculated from a first channel of at least two channels of a multichannel signal and a second channel of at least two channels of the multichannel signal. Both of these gain parameters, namely the side gain and the residual gain, are transmitted, stored, or generally output along with the downmix signal calculated from the multichannel signal by the downmixer.

[0008] Embodiments of the first aspect of the present invention are based on a novel mid / side technique that results in a new set of parameters, and in an encoder, a mid / side conversion is applied to the input channels to capture all the information from the two input channels together. The mid signal is the weighted average of the left and right channels, with weights that are complex and selected to compensate for phase differences. Thus, the side signal is the corresponding weighted difference of the input channels. Only the mid signal is waveform coded, while the side signal is parametrically modeled. The encoder operates in a subband, where the IPD and two gain parameters are extracted as stereo parameters. A first gain, called the side gain, is obtained from the prediction of the side signal by the mid signal, and a second gain, called the residual gain, captures the remaining energy relative to the energy of the mid signal. The mid signal then serves as a downmix signal sent to the decoder along with the stereo parameters.

[0009] The decoder estimates the lost side channel based on the side gain and residual gain, and then combines the two channels by using a substitute for the remainder.

[0010] The first aspect of the present invention is advantageous in that the side gain and the residual gain are limited to a specific small range of numbers. Specifically, the side gain is limited to a range of -1 to +1 in a preferred embodiment, and the residual gain is further limited to a range of 0 and 1. Even more useful in a preferred embodiment is that the residual gain depends on the side gain such that the range of values ​​the residual gain can have becomes smaller as the side gain increases.

[0011] Specifically, the side gain can be calculated as a side prediction gain applicable to the mid-signals of the first and second channels in order to predict the side signals of the first and second channels. The parameter calculator is also configured to calculate the residual gain as a residual prediction gain, which indicates the energy or amplitude of the residual signal of such prediction of the side signal by the mid-signal and the side gain.

[0012] However, importantly, it is not necessary to actually perform predictions on the encoder side, or to actually encode the side signals on the encoder side. Instead, the side gains and residual gains can be calculated using only amplitude-related measures, such as energy, power, or other characteristics related to the amplitudes of the left and right channels. In addition, the calculation of the side gains and residual gains relates only to the inner product between both channels; i.e., any other channels besides the left and right channels, such as the downmix channel itself or the side channel itself, do not need to be calculated in the embodiment. However, in other embodiments, the side signals may be calculated, various trials for prediction may be calculated, and gain parameters such as side gains and residual gains may be calculated from the residual signal associated with a particular side gain prediction, resulting in predefined criteria in various trials, such as the minimum energy of the residual or surplus signal. Thus, there is high flexibility and, nevertheless, low complexity for calculating the side gains on the one hand and the residual gains on the other.

[0013] Two advantages exist for the exemplary gain parameters over ILD and ICC. First, they naturally exist within a finite interval (side gain at [-1,1] and residual gain at [0,1]), in contrast to ILD parameters which can take any large or small value. Second, the calculation is simpler, as it involves only one special function evaluation, compared to two calculations for ILD and ICC.

[0014] A preferred embodiment of the first aspect relies on the calculation of parameters in the spectral domain, i.e., the parameters are calculated for various frequency bins, or more preferably for various subbands, each subband containing a certain number of frequency bins. In a preferred embodiment, the number of frequency bins contained within a subband increases from lower subbands to higher subbands to mimic the characteristics of human auditory perception, i.e., higher bands cover higher frequency ranges or bandwidths, and lower bands cover lower frequency ranges or bandwidths.

[0015] In a preferred embodiment, the downmixer calculates an absolutely phase-compensated downmix signal, where phase rotation is applied to the left and right channels based on the IPD parameters, but the phase compensation is performed so that channels with more energy are rotated less than channels with less energy. Side gains are preferably used to control the phase compensation, but in other embodiments, any other downmix may be used, which is also a special advantage of the present invention that a parametric representation of the side signal, i.e., side gain on the one hand and residual gain on the other, is calculated based only on the original first and second channels and does not require any information about the transmitted downmix. Thus, any downmix may be used with a new parametric representation consisting of side gains and residual gains, but the present invention is also particularly useful when applied in conjunction with absolute phase compensation based on side gains.

[0016] In a further embodiment of absolute phase compensation, the phase compensation parameter is the inverse tangent function (atan or tan) that arises when calculating the phase compensation parameter. -1The singularity of the function is specifically calculated based on a certain predetermined number so that it is moved from the center to a specific lateral position. This shift of the singularity ensures that no problems arise from the singularity for a phase shift of + / -180° and a gain parameter close to 0, i.e., for left and right channels having exactly the same energy. Such signals are known to occur fairly frequently, but signals that are out of phase with each other, for example, between 3dB and 12dB or with a difference of about 6dB, do not occur under natural circumstances. Thus the singularity is only shifted, but nevertheless, this shift has been found to improve the overall performance of the downmixer because it ensures that the singularity does not occur in signal configurations that occur far less frequently under normal circumstances than when a simple arctangent function has its singularity.

[0017] Further embodiments utilize the dependency of side gains and residual gains to implement an efficient quantization procedure. For this purpose, in the first embodiment, it is preferable to perform joint quantization, in which the side gains are quantized first, and then the residual gains are quantized using a quantization step based on the value of the side gains. However, other embodiments rely on joint quantization in which both parameters are quantized into a single code, and a particular portion of this code depends on a particular group of quantization points belonging to a particular level difference characteristic of the two channels encoded by the encoder.

[0018] A second aspect relates to an apparatus for downmixing a multichannel signal including at least two channels, the apparatus comprising: a downmixer for calculating a downmix signal from a multichannel signal, the downmixer being configured to calculate the downmix using absolute phase compensation such that, when calculating the downmix signal, only the channel having the lower energy of the at least two channels is rotated, or rotated more strongly than the channel having the higher energy; and an output interface for generating an output signal, the output signal including information about the downmix signal.

[0019] Preferably, rotation is performed on the minor channel, but in that case, it is possible in situations of small energy differences where the minor channel is not always rotated more than the major channel. However, when the energy ratio is sufficiently large or sufficiently small, a preferred embodiment rotates the minor channel more than the major channel. Therefore, preferably, the minor channel is rotated more than the major channel only when the energy difference is significant or exceeds a predefined threshold such as 1 dB or more. This applies to both downmixers and upmixers.

[0020] Preferred embodiments of the present invention will be discussed later with reference to the accompanying drawings. [Brief explanation of the drawing]

[0021] [Figure 1] This is a block diagram of a device for encoding a multi-channel signal according to one embodiment. [Figure 2] This is a block diagram of one embodiment of a parameter calculator. [Figure 3] This is a diagram of a further embodiment of the parameter calculator. [Figure 4] This is a diagram of one embodiment of a downmixer that performs absolute phase compensation. [Figure 5a]This is a block diagram of one embodiment of an output interface that performs a specific quantization. [Figure 5b] This diagram illustrates an example codeword. [Figure 6] This is a diagram of one embodiment of a device for decoding encoded multichannel signals. [Figure 7] This is a diagram of one embodiment of an upmixer. [Figure 8] This is a diagram of one embodiment of a residual signal combiner. [Figure 9] This is a diagram illustrating one embodiment of an input interface. [Figure 10a] This demonstrates how to handle duplicate frames. [Figure 10b] This figure shows one embodiment of a time-spectrum converter. [Figure 10c] This figure shows the spectrum of the left or right channel and the composition of various subbands. [Figure 10d] This figure shows one embodiment of a spectrum-time converter. [Figure 11] This figure shows the lines relating to conditional quantization in the first embodiment. [Figure 12] This figure shows lines relating to coupled quantization according to further embodiments. [Figure 13] This figure shows the coupled quantization points with respect to side gains and residual gains. [Modes for carrying out the invention]

[0022] Figure 1 shows a device for encoding a multichannel signal containing at least two channels. Specifically, the multichannel signal is shown as 100 in Figure 1, having a first channel 101 and a second channel 102, with no additional channels or having any selected number of additional channels, and further additional channels are shown as 103.

[0023] The multi-channel signal 100 is input to a downmixer 120 for calculating a downmix signal 122 from the multi-channel signal 100. Depending on the specific implementation, the downmixer may use the first channel 101, the second channel 102, and the third channel 103, or only the first and second channels, or all channels of the multi-channel signal to calculate the multi-channel signal.

[0024] Furthermore, the encoding apparatus includes a parameter calculator 140 for calculating side gains 141 from at least two channels, a first channel 101 and a second channel 102, and in addition, the parameter calculator 140 calculates residual gains 142 from the first and second channels. In other embodiments, an optional inter-channel phase difference (IPD) is also calculated, as shown in 143. The downmix signal 122, side gains 141, and residual gains 142 are transferred to an output interface 160 that generates an encoded multi-channel signal 162 containing information about the downmix signal 122, side gains 141, and residual gains 142.

[0025] It should be noted that side gains and residual gains are typically calculated per frame, such that a single side gain and a single residual gain are calculated for each frame. However, in other embodiments, not only are single side gains and single residual gains calculated per frame, but groups of side gains and groups of residual gains are calculated per frame, where each side gain and each residual gain relates to a particular subband of the first and second channels. Thus, in a preferred embodiment, the parameter calculator calculates groups of side gains and groups of residual gains for each frame of the first and second channels, where the number of side gains and residual gains for a frame is typically equal to the number of subbands. When a high-resolution time-spectral transform such as a DFT is applied, the side gains and residual gains for a particular subband are calculated from a group of frequency bins of the first and second channels. However, when a low-resolution time-frequency transform resulting in subband signals is applied, the parameter calculator 140 calculates the side gains and residual gains for each subband, or even for a group of subbands.

[0026] When side gains and residual gains are calculated for a group of subband signals, the parameter resolution is reduced, resulting in a lower bitrate but a lower quality representation of the parametric representation of the side signals. In other embodiments, the time resolution may also be modified so that side gains and residual gains are calculated for a group of frames rather than for each frame, where the group of frames has two or more frames. Thus, in such embodiments, it is preferable to calculate subband-related side / residual gains, where the side / residual gains are related to a group of frames that include two or more frames, but are related to a particular subband. Accordingly, according to the present invention, the time and frequency resolution of the parameter calculations performed by block 140 can be modified with high flexibility.

[0027] The parameter calculator 140 is preferably implemented as outlined in Figure 2 with respect to a first embodiment, or as outlined in Figure 3 with respect to a second embodiment. In the embodiment of Figure 2, the parameter calculator comprises a first time-spectrum converter 21 and a second time-spectrum converter 22. Furthermore, the parameter calculator 140 of Figure 1 comprises a calculator 23 for calculating a first amplitude-related characteristic, a calculator 24 for calculating a second amplitude-related characteristic, and a calculator 25 for calculating the inner product of the outputs of blocks 21 and 22, i.e., the inner product of the spectral representations of the first and second channels.

[0028] The outputs of blocks 23, 24, and 25 are forwarded to the side gain calculator 26 and also to the residual gain calculator 27. The side gain calculator 26 and the residual gain calculator 27 apply specific relationships between the first amplitude-related characteristics, the second amplitude-related characteristics, and the dot product, and the relationships applied by the residual gain calculator to combine both inputs are different from the relationships applied by the side gain calculator 26.

[0029] In a preferred embodiment, the first and second amplitude-related characteristics are the energy within the subband. However, other amplitude-related characteristics may relate to the amplitude in the subband itself, to the signal power within the subband, or to any other power of the amplitude with an exponent greater than 1, where the exponent can be a real number greater than 1, or an integer greater than 1, such as an integer of 2 relating to signal power and energy, or an integer of 3 relating to loudness. Thus, each amplitude-related characteristic can be used to calculate the side gain and residual gain.

[0030] In a preferred embodiment, the side gain calculator and the residual gain calculator 27 are configured to calculate the side gain as a side prediction gain applicable to the mid-signals of the first and second channels in order to predict the side signals of the first and second channels, or the parameter calculator, in particular the residual gain calculator 27, is configured to calculate the residual gain as a residual prediction gain that indicates an amplitude-related measure of the residual signal of the prediction of the side signal by the mid-signal using the side gain.

[0031] Specifically, the parameter calculator 140 and the side gain calculator 26 in Figure 2 are configured to calculate the side signal using a fraction having a numerator and a denominator, where the numerator includes the amplitude characteristics of the first and second channels, and the denominator includes the amplitude characteristics of the first and second channels and a value derived from the dot product. The value derived from the dot product is preferably the absolute value of the dot product, but alternatively it could be any power of the absolute value, such as a power greater than 1, or even a characteristic different from the absolute value, such as the complex conjugate term or the dot product itself.

[0032] In further embodiments, the parameter calculator, residual gain calculator 27 in Figure 2, also uses fractions having a numerator and denominator that use both the value derived from the dot product and other parameters. Again, the value derived from the dot product is preferably the absolute value of the dot product, but alternatively, it can be any power of the absolute value, such as a power greater than 1, or even a property different from the absolute value, such as the complex conjugate term or the dot product itself.

[0033] Specifically, the side gain calculator 26 in Figure 2 is configured to use the energy difference of the first channel to calculate the side gain, the denominator using the sum of the energy or amplitude characteristics of both channels, and in addition using the dot product, preferably twice the dot product, although other multipliers for the dot product may also be used.

[0034] The residual gain calculator 27 is configured to use a weighted sum of the amplitude characteristics of the first and second channels and an inner product in the numerator, wherein the inner product is subtracted from the weighted sum of the amplitude characteristics of the first and second channels. The denominator for calculating the residual gain calculator includes a sum of the amplitude characteristics of the first and second channels and the inner product, wherein the inner product is preferably multiplied by 2, but may also be multiplied by other similar coefficients.

[0035] Furthermore, as indicated by connection line 28, the residual gain calculator 27 is configured to calculate the residual gain using the side gain calculated by the side gain calculator.

[0036] In a preferred embodiment, the residual gain and the side gain operate as follows. Specifically, an inter-channel phase difference for each band, which will be described later, can be calculated. However, before specifically outlining the calculation of the side gain as will be shown later in formula (9) and the specific preferred calculation of the side gain as will be shown later in formula (10), further description of the encoder is provided, which also refers to the calculation of IPD and downmixing in addition to the calculation of gain parameters.

[0037] Encoding of stereo parameters and calculation of the downmix signal are performed in the frequency domain. For this purpose, time-frequency vectors L of the left and right channels t and R t are generated by simultaneously applying an analysis window followed by a discrete Fourier transform (DFT). The DFT bins are then grouped into subbands (L t ,k) K ∈I b , (R t ,k) K ∈I b , where I b represents a set of subband indexes.

[0038] Calculation of IPD and Downmixing

[0039] Regarding downmixing, the inter-channel phase difference (IPD) for each bandwidth is:

[0040]

number

[0041] It is calculated as follows, where z * This represents the complex conjugate of z. This is the case for k∈I b Regarding the mid and side signals for each bandwidth,

[0042]

number

[0043] and

[0044]

number

[0045] Used to generate the absolute phase rotation parameter β is

[0046]

number

[0047] Given by, where g t,b represents the side gain specified below, where atan2(y,x) is the two-argument inverse tangent function whose value is the angle between the point (x,y) and the positive x-axis. It is intended to perform IPD compensation on channels with rather less energy. The coefficient 2 is IPD t,b = ±π and g t,b The singularity at =0 is IPD t,b = ±π and g t,bShift it to -1 / 3. In this way, toggling of β is avoided in a phase-shifted situation where the left and right channels have nearly equal energy distributions. The downmix signal is then subjected to the inverse DFT M t It is generated by applying it to and continuing with the composite window and duplicate addition.

[0048] In other embodiments, other inverse tangent functions different from the atan2 function, such as a simple tangent function, may be used similarly, but the atan2 function is preferred due to its safe application to the problem raised.

[0049] Calculation of gain parameters

[0050] In addition to the IPD for each frequency band, two other stereo parameters are extracted. t,b by S t,b The optimal gain for predicting the remaining energy p t,k =S t,k -g t,b M t,k (5) A few grams that minimizes it t,b , and mid signal M t When applied, p in each band t and M t The gain coefficient r that makes the energy equal t,b , in other words,

[0051]

number

[0052] The optimal prediction gain is the energy in the subband.

[0053]

number

[0054] , and also, L t and R tThe absolute value of the dot product

[0055]

number

[0056] from,

[0057]

number

[0058] It can be calculated as follows: From this, g t,b This means that it lies within [-1,1]. The residual gain is calculated from the energy and the inner product,

[0059]

number

[0060] It can be calculated similarly, and this is,

[0061]

number

[0062] This means, t,b This shows that ∈[0,1]. Thus, stereo parameters can be calculated independently of the downmix by calculating the corresponding energy and inner product. Specifically, the residual p is used to calculate the energy. t,k There is no need to calculate E. The calculation of the gain involves only one special function evaluation, whereas E L,t,b , E R,t,b , and X L / R,t,b There are two ways to calculate ILD and ICC from this: namely, the square root and the logarithm.

[0063]

number

[0064] and

[0065]

number

[0066] It is noteworthy that it includes this.

[0067] Decreased parameter resolution

[0068] If a lower parameter resolution, such as that given by the window length, is desired, then in (9) and (10), X L / R,t,b of

[0069]

number

[0070] Replaced by E L,t,b , E R,t,b Each of them,

[0071]

number

[0072] The gain parameters over successive windows of h may be calculated by replacing with , where the side gain is the weighted average of the side gains for each window, where the weights are M t+i,k It depends on the energy of the band, or the energy E per band. M,s,b It depends on, where s is the sum of the indices in equations 14 and 15.

[0073] Similarly, the IPD value then spans several windows,

[0074]

number

[0075] The calculation is performed as follows. Preferably, the parameter calculator 140 shown in Figure 1 is configured to calculate the subband-by-subband representation as a sequence of complex-valued spectra, where each spectrum relates to a time frame of a first channel or a second channel, where the time frames of the sequence are adjacent to each other, and where the adjacent time frames overlap each other.

[0076] Furthermore, the parameter calculator 140 is configured to calculate the first and second amplitude-related measures by squaring the magnitude of the complex spectral values ​​within the subbands and summing the squared magnitudes within the subbands, as previously shown in equation (7), for example, where index b represents the subband.

[0077] Furthermore, as outlined in Equation 8, the parameter calculator 140, and in particular the inner product calculator 25 in Figure 2, is configured to calculate the inner product by summing products within subbands, where each product includes the spectral value of the first channel at the frequency bin and the shared complex spectral value of the second channel with respect to the frequency bin. The magnitude of the sum is then formed.

[0078] As outlined in Equations 1 to 4, it is preferable to use absolute phase compensation. Therefore, in this embodiment, the downmixer 120 is configured to calculate the downmix 122 using absolute phase compensation such that when calculating the downmix signal, only the channel with the lower energy of the two channels is rotated, or the channel with the lower energy of the two channels is rotated more strongly than the other channel with the higher energy. Such a downmixer 120 is shown in Figure 4. Specifically, the downmixer comprises an inter-channel phase difference (IPD) calculator 30, an absolute phase rotation calculator 32, a downmix calculator 34, and an energy difference or side gain calculator 36. It should be emphasized that the energy difference or side gain calculator 36 may be implemented as the side gain calculator 26 in Figure 2. However, alternatively, there may be different implementations in block 36 that calculate only the difference in an energy difference, or more generally, the difference in an amplitude-related characteristic which may be energy, power, or amplitude itself, or a power of amplitude that is added when the power is not 2, such as a power between one and two or more.

[0079] Specifically, exponents or powers of 3 correspond to loudness, for example, rather than energy.

[0080] Specifically, the IPD calculator 30 in Figure 4 is typically configured to calculate the inter-channel phase difference for each subband of each of the multiple subbands of the first and second channels 101 and 102 that are input to block 30. Furthermore, the downmixer has an absolute phase rotation parameter for each of the multiple subbands, again typically, which operates based on the energy difference provided by block 36 between the first and second channels, or generally based on the difference in amplitude-related characteristics between both channels 101 and 102. In addition, the downmix calculator 34 is configured to weight the first and second channels using the IPD parameter and the absolute phase rotation parameter, denoted as β, when calculating the downmix signal.

[0081] Preferably, block 36 is implemented as a side gain calculator so that the absolute phase rotation calculator operates based on the side gains.

[0082] Therefore, in a preferred embodiment, block 30 in Figure 4 is configured to implement formula (1), block 32 is configured to implement formula (4), and block 34 is configured to implement formula (2).

[0083] Specifically, side gain g t,bThe coefficient 2 in equation (4) preceding the term containing can be set differently from 2, for example, preferably a value between 0.1 and 100. Naturally, -0.1 and -100 can also be used. This value ensures that the singularity present at an IPD of ±180° for approximately equal left and right channels is moved to a different location, i.e., a different side gain of -1 / 3 for coefficient 2. However, other coefficients different from 2 can be used. These other coefficients move the singularity to a different side gain parameter of -1 / 3. All these different coefficients have been found useful because they achieve that the singularity in question is in a “location” within the soundstage with the relevant left and right channel signals that occur less frequently than signals that are typically out of phase and have equal or approximately equal energy.

[0084] In a preferred embodiment, the output interface 160 in Figure 1 is configured to perform parametric information quantization, i.e., quantization of side gains, such as those provided on line 141 by the parameter calculator 140, and residual gains, such as those provided on line 142 from the parameter calculator 140 in Figure 1.

[0085] Specifically, in embodiments where the residual gain depends on the side gain, if it is preferable to quantize the side gain and then the residual gain, then in this embodiment, the quantization step for the residual gain depends on the value of the side gain.

[0086] Specifically, this is shown in Figure 11, and similarly in Figures 12 and 13.

[0087] Figure 1 shows the line relating to conditional quantization. Specifically, the residual gain is always (1-g 2 ) 1 / 2It has been shown that it is within the range determined by g. Therefore, when g=0, r can be in the range between 0 and 1. However, when g is equal to 0.5, r can be in the range between 0.866 and 0. Furthermore, for example when g=0.75, the range r is restricted to between 0 and 0.66. In the extreme embodiment of g=0.9, r can only be in the range between 0 and 0.43. Furthermore, when g=0.99, r can only be in the range between 0 and 0.14.

[0088] Therefore, this dependency can be used to reduce the quantization step size of the residual gain quantization for higher side gains. Thus, considering Figure 11, the vertical lines showing the range of r values ​​can always be divided by a certain integer such as 8, so that each line has 8 quantization steps. Thus, it is clear that for lines reflecting higher side gains, the quantization steps are smaller than for lines with lower side gains. Therefore, higher side gains can be quantized more finely without any increase in bit rate.

[0089] In a further embodiment, the quantizer is configured to perform coupled quantization using groups of quantization points, where each group of quantization points is defined by a fixed amplitude correlation ratio between a first channel and a second channel. An example of an amplitude correlation ratio is the energy between left and right, i.e., this means the line with respect to the same ILD between the first channel and the second channel, as shown in Figure 12. In this embodiment, the output interface is configured as shown in Figure 5a and comprises a subband-by-subband ILD calculator that receives the first channel and the second channel, or alternatively, side gains g and residual gains r, as inputs. The subband-by-subband ILD calculator, indicated by reference no. 50, outputs a specific ILD with respect to the parameter values ​​g, r to be quantized. The ILD, or generally the amplitude correlation ratio, is transferred to a group matcher 52. The group matcher 52 determines the most matching group and transfers this information to a point matcher 54. Both the group matcher 52 and the point matcher 54 feed into a code builder 56, which ultimately outputs a code, such as a codeword from a codebook.

[0090] Specifically, the code builder receives the code for the side gain g and determines the code bit 57a shown in Figure 5b, which indicates the code for g and r for the subband. Furthermore, the group matcher, having determined a specific group of quantization points that match the determined ILD, outputs bits 2 through 5 shown in 57b as the group ID. Finally, the point matcher outputs bits 6 through 8 in the embodiment shown in Figure 5b, 57c, where these bits indicate the point ID, i.e., the ID of the quantization point within the group indicated by bit 57b. Figure 5b shows an 8-bit code with a single code bit, four group bits, and three point bits, but other codes with a code bit, more or fewer group bits, and more or fewer point bits may be used. Due to the fact that side gains have both positive and negative values, the group bits and point bits, i.e., set of bits 57b and set of bits 57c, have only either purely negative values ​​or preferably purely positive values, and if the sign bit shows a negative sign, the residual gain is always decoded as a positive value, but the side gain is decoded as a negative value, which means that the energy of the left channel is lower than the energy of the right channel when the rule shown in Equation 9 is applied to calculate the side gain.

[0091] Next, further embodiments relating to quantization will be outlined.

[0092] Quantization of side gains and residual gains

[0093] The inequality in (11) reveals the strong dependence of the residual gain on the side gain, since the side gain determines the range of the residual gain. Therefore, when g tends toward ±1, the number of possible quantization points for r decreases, so independently quantizing the side gain g and residual gain r by selecting quantization points within [-1,1] and [0,1] is inefficient.

[0094] Conditional quantization

[0095] There are various ways to handle this problem. The simplest way is to first quantize g, and then the quantization point is an interval.

[0096]

number

[0097] The quantized values ​​that will be included

[0098]

number

[0099] The condition is to quantize r. The quantization points can then be uniformly selected along these quantization lines, some of which are shown in Figure 11.

[0100] joint quantization

[0101] A more sophisticated method for selecting quantization points is to examine the lines in the (g,r) plane corresponding to the fixed energy ratio between L and R. 2 If ≥1 exhibits such an energy ratio, the corresponding line is (0,s) for 0≦s≦1 when c=1, or

[0102]

number

[0103] It is given by either of the following. This is L t and R t It is possible to exchange g t,b Only the sign of r is changed, t,b Since it will not be changed, c 2 It also covers the case where <1.

[0104] This method covers larger regions with the same number of quantization points, as can be seen from Figure 12. Again, the quantization points on a line can be selected uniformly, for example, according to the length of each line. Other possibilities include selecting them to match a pre-selected ICC value, or optimizing them acoustically.

[0105] The quantization method that has been found to work well is the ILD value. ±{0,2,4,6,8,10,13,16,19,22,25,30,35,40,45,50} (23) Based on the corresponding energy lines, eight quantization points are selected for each. This results in a codebook with 256 entries, organized as an 8x16 table of quantization points, each holding a value and sign bit corresponding to non-negative values ​​of g. This results in an 8-bit integer representation of the quantization point (g,r), where, for example, the first bit specifies the sign of g, the next four bits hold the column index in the 8x16 table, and the last three bits hold the row index.

[0106] (g t,b ,r t,b The quantization of ) can be performed by a thorough codebook search, but it is more efficient to first calculate the subband ILD and restrict the search to the most matching energy line. Thus, only eight points need to be considered.

[0107] Inverse quantization is performed by a simple table lookup.

[0108] Figure 12 shows the 128 quantization points for this scheme that cover non-negative values ​​of g.

[0109] While procedures are disclosed for calculating the side gain and residual gain without actually calculating the side signal, i.e., the signal difference between the left signal and the right signal as shown in equations (9) and (10), further embodiments operate to calculate the side gain and residual gain differently, i.e., using the actual calculation of the side signal. This procedure is shown in Figure 3.

[0110] In this embodiment, the parameter calculator 140 shown in Figure 1 includes a side signal calculator 60, which receives a first channel 101 and a second channel 102 as inputs and outputs an actual side signal that may be calculated in the time domain, but preferably in the frequency domain, as shown by, for example, Equation 3. However, although Equation 3 shows the calculation of the side signal using the absolute phase rotation parameter β and IPD parameter on a band and frame-by-band and frame-by-frame basis, the side signal can also be calculated without phase compensation. Equation 3 is L t,k and R t,k This results in an equation that yields only this. Therefore, the side signal can also be calculated as a simple difference between the left and right channels or between the first and second channels, and normalization using the square root of 2 may or may not be used.

[0111] The side signal calculated by the side signal calculator 60 is transferred to the residual signal calculator 61. The residual signal calculator 61 performs the procedure shown in Equation 5, for example. The residual signal calculator 61 calculates different test side gains, i.e., side gain g d,b The system is configured to use different values ​​for the testside gain, i.e., different testside gains for the same bandwidth and frame, resulting in different residual signals being obtained, as shown by the multiple outputs of block 61.

[0112] The side gain selector 62 in Figure 3 receives all different residual signals and selects a test side gain associated with one of the different residual signals, or one of the different residual signals that satisfies a predefined condition. This predefined condition may be, for example, selecting a side gain that results in the residual signal with the least energy among all the different residual signals. However, other predetermined conditions may be used, such as the smallest amplitude-related condition, which is different from energy, such as loudness. However, other procedures may also be applied, such as using a residual signal with an energy that is one of the five smallest energies, rather than the smallest energy. In practice, the predefined condition may also be selecting a residual signal that exhibits certain other audio characteristics, such as certain features within a particular frequency range.

[0113] The selected specific test side gain is determined by the side gain selector 62 as a side gain parameter for a specific frame or a specific bandwidth and a specific frame. The selected residual signal is transferred to the residual gain calculator 63, which in one embodiment can simply calculate the amplitude-related characteristics of the selected residual signal, or, preferably, can calculate the residual gain as a relationship between the amplitude-related characteristics of the residual signal and the amplitude-related characteristics of the downmix signal or mid signal. Even if a downmix is ​​used that is different from a phase-compensated downmix or a downmix consisting of a sum of left and right signals, the residual gain may, in some cases, be related to the unphase-compensated sum of left and right signals.

[0114] Therefore, while Figure 3 shows a method for calculating the side gain and residual gain using actual side signal calculations, in the embodiment of Figure 2, which largely reflects Equations 9 and 10, the side gain and residual gain are calculated without explicit calculation of the side signal and without performing residual signal calculations using different test side gains. Thus, both embodiments yield side gain and residual gain that parameterize the residual signal from predictions, and it becomes clear that other procedures are also possible for calculating side gain and residual gain other than those shown in Figures 2 and 3 or by the corresponding Equations 5 to 10.

[0115] Furthermore, it should be noted that all given formulas are always preferred embodiments for the values ​​determined by the corresponding formulas. However, while the benefits increase with smaller deviations from the values ​​determined by the formulas, values ​​that differ from the values ​​determined by the corresponding formulas, preferably within a range of ±20%, are also useful and have been found to already offer advantages over the prior art. Therefore, in other embodiments, it is preferable to use values ​​that differ by only ±10% from the values ​​determined by the corresponding formulas, and in the most preferred embodiment, the values ​​determined by the formulas are values ​​used for the calculation of several data items.

[0116] Figure 6 shows a device for decoding the encoded multichannel signal 200. The decoding device comprises an input interface 204, a residual signal combiner 208 connected to the input interface 204, and an upmixer 212 connected to the input interface 204 on one side and to the residual signal combiner 208 on the other. In a preferred embodiment, the decoder further comprises a spectrum-time converter 216 to ultimately output first and second channels in the time domain, as shown in 217 and 218.

[0117] Specifically, the input interface 204 is configured to receive the encoded multichannel signal 200 and to obtain a downmix signal 207, a side gain g 206, and a residual gain r 205 from the encoded multichannel signal 200. The residual signal combiner 208 is configured to combine the residual signals using the residual gain 205, and the upmixer 212 is configured to upmix the downmix signal 207 using the side gain 206 and the residual signal 209 determined by the residual signal combiner 208 in order to obtain a reconstructed first channel 213 and a reconstructed second channel 214. In embodiments where the residual signal combiner 208 and the upmixer 212 operate in the spectral domain, or where at least the upmixer 212 operates in the spectral domain, the reconstructed first and second channels 213, 214 are given in spectral domain representations, and the spectral domain representation for each channel can be converted to the time domain by the spectral-to-time converter 216 to ultimately output the first and second reconstructed channels in the time domain.

[0118] Specifically, the upmixer 212 is configured to perform a first weighting operation using the first weighter 70 shown in Figure 7 to obtain a first weighted downmix channel. Furthermore, the upmixer performs a second weighting operation again using the second weighter, with the side gain 206 on one hand and the downmix signal 207 on the other, to obtain a second weighted downmix signal. Preferably, the first weighting operation performed by block 70 is different from the second weighting operation performed by block 71, such that the first weighted downmix 76 is different from the second weighted downmix 77. Furthermore, the upmixer 212 is configured to calculate the reconstructed first channel using the combination of the first weighted downmix signal 76 and the residual signal 209 performed by the first combiner 72. Furthermore, the upmixer also includes a second combiner 73 for performing a second combination of a second weighted downmix signal 77 and a residual signal 209.

[0119] Preferably, the combination rules performed by the first combiner 72 and the second combiner 73 are different from each other, such that the output of block 72 and the output of block 73 are substantially different from each other, due to the different combination rules in blocks 72 and 73, and the different weighting rules performed by blocks 70 and 71.

[0120] Preferably, the first and second combination rules are different from each other due to the fact that one combination rule is an addition operation and the other combination rule is a subtraction operation. However, other pairs of the first and second combination rules can be used in the same way.

[0121] Furthermore, since one weighting rule uses weighting determined by a weighting coefficient determined by the difference between a predetermined number and the side gain, and the other weighting rule uses a weighting coefficient determined by the sum between a predetermined number and the side gain, the weighting rules used in blocks 70 and 71 are different from each other. The predetermined number may be equal to or different from each other in both weighters, and the predetermined number may be an integer or a non-integer, not zero, and preferably equal to 1.

[0122] Figure 8 shows a preferred implementation of the residual signal combiner 208. The residual signal combiner 208 comprises a kind of raw residual signal selector, or generally, an uncorrelated signal calculator 80. Furthermore, the signal output by block 80 is input to a weighter 82 which receives the residual gain output by the input interface 204 in Figure 6, indicated by reference no. 205, as input. The residual signal combiner also preferably comprises a normalizer 84, which receives the mid-signal 85 of the current frame as input and the signal output by block 80, i.e., the raw signal or uncorrelated signal 86, as further input. Based on these two signals, a normalization coefficient g is calculated. norm87 is calculated, where the normalization coefficient 87 is preferably used by the weighter 82 together with the residual gain r to finally obtain the synthesized residual signal 209.

[0123] In a preferred embodiment, the raw residual signal selector 80 is configured to select a downmix signal of a preceding frame, such as the immediately preceding frame or an even earlier frame. However, depending on the implementation, the raw residual signal selector 80 may be configured to select a left or right signal or a first or second channel signal calculated for the preceding frame, or the raw residual signal selector 80 may determine the residual signal based on a combination such as the sum, difference, etc., of the left and right signals determined for either the immediately preceding frame or an even earlier preceding frame. In other embodiments, the uncorrelated signal calculator 80 may also be configured to actually generate an uncorrelated signal. However, the raw residual signal selector 80 operates without specific uncorrelated filters such as reverberation filters, but for reasons of low complexity, it is preferable to select only signals that already exist from the past, such as the mid signal, reconstructed left signal, reconstructed right signal, or weighted combinations that do not depend on specific reverberation or uncorrelated filters, i.e., signals derived from earlier reconstructed left and right signals by simple operations such as (weighted) addition, (weighted) subtraction, etc.

[0124] Generally, the weighter 82 is configured to calculate the residual signal such that the energy of the residual signal is equal to the signal energy indicated by the residual gain r, where this energy may be shown in absolute terms, but preferably in terms relative to the mid signal 85 of the current frame.

[0125] In preferred embodiments relating to the encoder and decoder sides, the side gain values, and appropriately the values ​​from the residual gain, are not zero.

[0126] Next, an additional preferred embodiment relating to the decoder is given in the form of an equation.

[0127] Upmixing is performed again in the frequency domain. For this purpose, the time-frequency conversion from the encoder is applied to the decoded downmix, and the time-frequency vector is used.

[0128]

number

[0129] This brings about the inversely quantized value.

[0130]

number

[0131] ,

[0132]

number

[0133] , and

[0134]

number

[0135] Using this, the left and right channels are k∈I b In contrast,

[0136]

number

[0137] and

[0138]

number

[0139] It is calculated as follows, and here,

[0140]

number

[0141] This is the missing residual ρ from the encoder. t,k It is a substitute for g norm This is the energy regulation coefficient.

[0142]

number

[0143] Therefore, this energy adjustment coefficient is the relative gain coefficient.

[0144]

number

[0145] To make something absolute. For example,

[0146]

number

[0147] We can take, and here, d b >0 represents the frame delay per bandwidth. Phase rotation coefficient

[0148]

number

[0149] teeth,

[0150]

number

[0151] It is then recalculated as follows. The left and right channels are then subjected to the inverse DFT.

[0152]

number

[0153] and

[0154]

number

[0155] It is generated by applying it to and continuing with the composite window and duplicate addition.

[0156] Figure 9 shows a further embodiment of the input interface 204. This embodiment reflects the inverse quantization operation, as previously described with respect to the encoder side with respect to Figures 5a and 5b. Specifically, the input interface 204 includes an extractor 90 that extracts a coupled code from an encoded multichannel signal. This coupled code 91 is transferred to a coupled codebook 92, which is configured to output, for each code, code information, group information, or point information, or to output, for each code, the final inverse quantization value g and the final inverse quantization value r, i.e., the inverse quantization side gain and the inverse quantization residual gain.

[0157] Figure 10a shows a schematic representation of the first and second channels, or left and right channels l(t) and r(t), in the time domain.

[0158] In embodiments where side gains and residual gains are calculated in the spectral domain, the left and right channels, or the first and second channels, are preferably separated into overlapping frames F(1), F(2), F(3), and F(4), etc. In the embodiment shown in Figure 10a, the frames overlap by 50%, but other overlaps are equally useful. Furthermore, only two-frame overlaps are shown, i.e., it is shown that only two subsequent frames always overlap each other. However, multiple overlapping frames, such as three, four, or five overlapping frames, can also be used. In this case, the advance value, i.e., how much the next frame differs from the current frame, is just smaller, such as 10%, 20%, or 30%, rather than 50%, as in the embodiment shown in Figure 10a.

[0159] Figure 10b shows a preferred implementation of a time-spectrum converter, such as block 21 or block 22 shown in Figure 2. Such a time-frequency converter receives a sequence of frames l(t) or r(t) as input. The analysis windower 1300 then outputs a sequence of windowed frames, all of which are windowed using preferably the same analysis window. The analysis window can be a sine window or any other window, and a separate sequence is calculated for the first channel, and yet another separate sequence is calculated for the second channel.

[0160] The windowed frame sequence is then input to the transformation block 1302. Preferably, the transformation block 1302 performs a transformation algorithm that yields complex spectral values, such as a DFT, specifically an FFT. However, in other embodiments, a purely real transformation algorithm such as a DCT or MDCT (Modified Discrete Cosine Transform) may also be used, after which the imaginary part may be estimated from the purely real part, as is known in the art, for example, as implemented in the USAC (Integrated Speech and Audio Coding) standard. Other transformation algorithms may be subband filter banks, such as a QMF filter bank, that yield complex-valued subband signals. Typically, the subband signal filter bandwidth has a lower frequency resolution than the FFT algorithm, and an FFT or DFT spectrum with a certain number of DFT bins may be transformed into a subband-by-subband representation by collecting a specific number of bins. This is shown in Figure 10c.

[0161] Specifically, Figure 10c shows the first or second channel L for a particular frame t. k , R k The complex spectrum of the frequency domain representation is shown. Spectral values ​​are given in magnitude / phase representation or in real / imaginary part representation. Typically, the DFT yields frequency bins having the same frequency resolution or bandwidth. However, preferably, side gains and residual gains are calculated per subband to reduce the number of bits required to transmit residual and side gains. Preferably, the subband representation is generated using subbands that increase from lower frequencies to higher frequencies. Thus, in one example, subband 1 may have a first number of frequency bins, such as 2 bins, and second higher subbands, such as subband 2, subband 3, or any other subband, may have a larger number of frequency bins, such as 8 frequency bins, as shown by subband 3. Thus, the frequency bandwidths of individual subbands may preferably be tuned to the characteristics of the human ear, as is known in the art with respect to the Burke scale.

[0162] Therefore, Figure 10c shows the different frequency bins indicated by the parameter k in the previously disclosed equation, and the individual subbands shown in Figure 10c are indicated by the subband index b.

[0163] Figure 10d shows an implementation of the spectrum-time converter, for example, as implemented by block 216 in Figure 6. The spectrum-time converter requires a backward transformer 1310, followed by a composite windower 1312, and then a repeater / adder 1314 to ultimately obtain time-domain channels. Thus, at the input to 1310 are the reconstructed spectral-domain channels 213 and 214 shown in Figure 6, and at the output of the repeater / adder 1340 are the reconstructed first and second time-domain channels 217 and 218.

[0164] The inverse converter 1310 is configured to perform an algorithm that results in the inverse transform, specifically, preferably the reverse algorithm of the algorithm applied in block 1302 of Figure 10b on the encoder side. Furthermore, the composite windower 1312 is configured to apply a composite window that matches the corresponding analysis window, preferably the same analysis and composite windows are used, but this is not necessarily the case. The duplicate adder 1314 is configured to perform duplicates as shown in Figure 10a. Thus, the duplicate / adder 1314 takes, for example, the composite windowed frame corresponding to F(3) in Figure 10a, in addition to the composite windowed frame F(4) in Figure 10a, and then, sample by sample, adds the corresponding samples of the second half of F(3) to the corresponding samples of the first half of F(4) in order to finally obtain the samples of the actual time-domain output channel.

[0165] Next, different specific embodiments of the present invention are briefly given.

[0166] • Stereo M / S with IPD compensation and absolute phase compensation according to equation (4). • Stereo M / S by IPD compensation and prediction of S by M by (10) Stereo M / S ratio obtained by IPD compensation, prediction of S by M using (9), and residual prediction using gain coefficient (10). • Efficient quantization of side gain coefficients and residual gain coefficients through coupled quantization L in the (g,r) plane t and R t Coupled quantization of the side gain coefficients and residual gain coefficients on the line corresponding to the fixed energy ratio.

[0167] Preferably, it should be noted that all five different embodiments described above are implemented within the same encoder / decoder framework. However, it should be further noted that each of the embodiments given above can also be implemented separately from one another. Thus, the first embodiment using IPD compensation and absolute phase compensation can be implemented in any downmixer, regardless of any side gain / residual gain calculations. Furthermore, for example, the embodiments of side gain calculation and residual gain calculation can also be implemented using any downmix, i.e., using a downmix that is not calculated by a particular phase compensation.

[0168] Furthermore, the calculation of side gains on the one hand and the calculation of residual gains on the other hand can be performed independently of each other, and the calculation of side gains alone or together with any other parameter different from the residual gains is also advantageous to the art in that field, particularly with respect to ICC or ILD calculations, and furthermore, the calculation of residual gains alone or together with any other parameter different from the side gains is also already useful.

[0169] Furthermore, efficient coupled quantization or conditional quantization of side gains and residual gains or gain coefficients is useful for any particular downmix. Thus, efficient quantization can also be used without any downmix at all. And this efficient quantization can be applied to any other parameter whose second parameter depends on the first parameter with respect to its range of values, so that a very efficient quantization of low complexity can be performed for such dependent parameters, which of course may also be different parameters from the side gains and residual gains.

[0170] Therefore, all five embodiments described above may be performed and implemented independently or together in a particular encoder / decoder implementation, and only subgroups of embodiments may be implemented together, that is, in some cases three embodiments may be implemented together without the other two embodiments, or only two of the five embodiments may be implemented together without the other three embodiments.

[0171] While some embodiments have been described in the context of the apparatus, it is clear that these embodiments also represent descriptions of the corresponding methods, where a block or device corresponds to a method step or a feature of a method step. Similarly, embodiments described in the context of a method step also represent descriptions of the corresponding blocks, items, or features of the corresponding apparatus.

[0172] Depending on specific implementation requirements, embodiments of the present invention may be implemented in hardware or software. Implementations may be carried out using digital storage media, such as floppy disks, DVDs, CDs, ROMs, PROMs, EPROMs, EEPROMs, or flash memory, which store electronically readable control signals that cooperate (or can cooperate) with a programmable computer system to perform the respective methods.

[0173] Some embodiments of the present invention include a data carrier having an electronically readable control signal that can cooperate with a programmable computer system so that one of the methods described herein is performed.

[0174] Generally, embodiments of the present invention may be implemented as a computer program product having program code, the program code being operable to perform one of the methods when the computer program product is executed on a computer. The program code may be stored, for example, on a machine-readable carrier.

[0175] Other embodiments include a computer program stored on a machine-readable carrier or non-temporary storage medium for performing one of the methods described herein.

[0176] In other words, one embodiment of the method of the present invention is a computer program having program code for performing one of the methods described herein when the computer program is executed on a computer.

[0177] Therefore, a further embodiment of the method of the present invention is a data carrier (or digital storage medium, or computer-readable medium) on which a computer program for performing one of the methods described herein is recorded.

[0178] Therefore, a further embodiment of the method of the present invention is a data stream or sequence of signals representing a computer program for performing one of the methods described herein. The data stream or sequence of signals may be configured to be transmitted, for example, over a data communication connection, for example, over the Internet.

[0179] Further embodiments include processing means configured or adapted to perform one of the methods described herein, such as a computer or a programmable logic device.

[0180] Further embodiments include a computer on which a computer program for performing one of the methods described herein is installed.

[0181] In some embodiments, a programmable logic device (e.g., a field-programmable gate array) may be used to perform some or all of the functions of the methods described herein. In some embodiments, the field-programmable gate array may work with a microprocessor to perform one of the methods described herein. Generally, the methods are preferably performed by any hardware device.

[0182] The embodiments described above are merely examples relating to the principles of the present invention. Modifications and variations of the configurations and details described herein will be obvious to those skilled in the art. Therefore, it is intended that the invention is limited only by the immediate claims and not by the specific details presented in the description and explanation of the embodiments herein.

[0183] References MPEG-4 High Efficiency Advanced Audio Coding (HE-AAC) v2 FROM JOINT STEREO TO SPATIAL AUDIO CODING-RECENT PROGRESS AND STANDARDIZATION, Proc. of the 7th International Conference on Digital Audio Effects (DAFX-04), Naples, Italy, October 5-8, 2004.

[0184] Furthermore, this specification includes embodiments as defined in the following sections (clauses). [Section 1] A device for encoding a multichannel signal (100) including at least two channels (101, 102), A downmixer (120) is used to calculate a downmix signal (122) from the multi-channel signal (100), A parameter calculator (140) for calculating side gains (141) from the first channel (101) and the second channel (102) of the at least two channels, and for calculating residual gains (142) from the first channel (101) and the second channel (102), An output interface (160) for generating an output signal, wherein the output signal includes information relating to the downmix signal (122) and information relating to the side gain (141) and the residual gain (142). A device equipped with the following features. [Section 2] The parameter calculator (140) A representation of each subband of the first channel and the second channel is generated (21), The first amplitude-related characteristics of the first channel in the subband are calculated (21, 22, 23, 24), and the second amplitude-related characteristics of the second channel in the subband are calculated. The dot product of the first channel and the second channel in the subband is calculated (25), The side gain in the subband is calculated using the first relationship including the first amplitude-related characteristic, the second amplitude-related characteristic, and the dot product (26) The residual gain in the subband is calculated using a second relationship that includes the first amplitude-related characteristic, the second amplitude-related characteristic, and the dot product (27). The second relationship is different from the first relationship, The apparatus according to claim 1, wherein the amplitude-related characteristics are determined from the amplitude, from the power, from the energy, or from any power of the amplitude having an exponent greater than 1. [Section 3] The apparatus according to claim 1 or 2, wherein the parameter calculator (140) is configured to calculate the side gain (141) and the residual gain (142) for each subband of a plurality of subbands of the first channel and the second channel. [Section 4] The parameter calculator is configured to calculate the side gain as a side prediction gain applicable to the mid-signals of the first and second channels in order to predict the side signals of the first and second channels, or The parameter calculator (140) is configured to use the side gain to calculate the residual gain as a residual prediction gain that shows the amplitude-related characteristics of the residual signal of the prediction of the side signal by the mid signal. The apparatus according to any one of claims 1 to 3. [Section 5] The parameter calculator (140) is configured to calculate the side gain (26) using a fraction having a numerator and a denominator, wherein the numerator includes the amplitude-related characteristics of the first channel and the amplitude-related characteristics of the second channel, and the denominator includes the amplitude-related characteristics of the first channel, the amplitude-related characteristics of the second channel and the dot product, or The parameter calculator (140) is configured to calculate the residual gain (27) using a fraction having a numerator and a denominator, wherein the numerator includes the dot product and the denominator includes the dot product. The apparatus according to any one of claims 1 to 4. [Section 6] The parameter calculator (140) is configured to calculate the side gain (26), wherein the numerator includes the difference between the first amplitude-related characteristic of the first channel and the second amplitude-related characteristic of the second channel, and the denominator includes the sum of the first amplitude-related characteristic of the first channel, the second amplitude-related characteristic of the second channel, and the value derived from the dot product, or The parameter calculator (140) is configured to calculate the residual gain (27) using the fraction having the numerator and the denominator, wherein the numerator includes the difference between the weighted sum of the first amplitude-related characteristic of the first channel and the second amplitude-related characteristic of the second channel and the dot product, and the denominator includes the sum of the amplitude-related characteristic of the first channel, the amplitude-related characteristic of the second channel and the value derived from the dot product. The apparatus according to claim 5. [Section 7] The apparatus according to any one of claims 1 to 6, wherein the parameter calculator (140) is configured to calculate the side gain with respect to a subband and (28) calculate the residual gain with respect to the subband using the side gain of the subband. [Section 8] The parameter calculator (140) determines that the value relating to the side gain is given by the following formula:

number

number

number

number

number

number

number

number

number

number

number

number

number

number

number

number

[0185] 21 First time-spectrum converter, block 22 Second time-spectral converter, block 23 Calculators, Blocks 24 Calculators, Blocks 25 Calculators, Inner Product Calculators, Blocks 26 Side Gain Calculator 27 Residual Gain Calculator 28 connecting lines 30-channel inter-channel phase difference (IPD) calculator, IPD calculator, block 32. Absolute Phase Rotation Calculator, Block 34 Downmix Calculator, Block 36 Energy difference or side gain calculator, block ILD calculator for each of the 50 subbands 52 Group Matching Unit 54 point matching box 56 Code Builder 57a Sign bit 57b Bits 2 through 5, set of bits 57c Bits 6 to 8, bit set 60 Side Signal Calculator 61 Residual signal calculator, block 62 Side Gain Selector 63 Residual Gain Calculator 70 First weighting device, block 71 blocks 72. First combiner, block 73. Second combiner, block 76 First weighted downmix, first weighted downmix signal 77 Second weighted downmix, second weighted downmix signal 80 Uncorrelated signal calculator, block, uncorrelated signal calculator 82 Weighting device 84 Normalizer 85 Mid-frame signal 86. Raw signal or uncorrelated signal 87 Normalization coefficient g norm , normalization coefficient 90 Extractor 91 Combined code 92 Combined Codebook 100 multi-channel signals 101 First Channel 102 Second channel 103 Additional channel, third channel 120 Down Mixer 122 Downmix signal, downmix 140 Parameter Calculator 141 Side Gain 142 Residual Gain 143 Inter-channel phase difference (IPD) 160 output interfaces 162 Encoded Multichannel Signal 200 Encoded Multichannel Signal 204 Input Interfaces 205 Residual gain r, Residual gain 206 Side Gain g, Side Gain 207 Downmix signal 208 Residual signal synthesizer 209 Residual signal 212 Up Mixer 213 First channel, spectral region channel 214 Second channel, spectral domain channel 216 Spectrum-Time Converter 217 First Channel 218 Second Channel 1300 Analytical Windowing Device 1302 conversion block, block 1310 Inverse Converter 1312 Synthetic windowizer 1314 Adder / Adder, Adder

Claims

1. A device for encoding a multichannel signal (100) including at least two channels (101, 102), A downmixer (120) is used to calculate a downmix signal (122) from the multi-channel signal (100), A parameter calculator (140) for calculating side gains (141) from the first channel (101) of the at least two channels (101, 102) and the second channel (102) of the at least two channels (101, 102), and for calculating residual gains (142) from the first channel (101) and the second channel (102), An output interface (160) for generating an output signal, wherein the output signal includes information relating to the downmix signal (122), information relating to the side gain (141), and information relating to the residual gain (142). Equipped with, The side gain (141) represents a side prediction gain applicable to the mid-signals of the first channel (101) and the second channel (102) in order to predict the side signals of the first channel (101) and the second channel (102), and the residual gain (142) represents a residual prediction gain that shows the amplitude-related characteristics of the residual signal of the prediction of the side signal by the mid-signal using the side gain (141). The parameter calculator (140) is configured to calculate the side gain (141) using a first fraction having a first numerator and a first denominator (26), wherein the first numerator includes a first amplitude-related characteristic of the first channel (101) and a second amplitude-related characteristic of the second channel (102), and the first denominator includes the first amplitude-related characteristic of the first channel (101), the second amplitude-related characteristic of the second channel (102), and the inner product of the first channel (101) and the second channel (102) in the subband, or The parameter calculator (140) is configured to calculate the residual gain (142) using a second fraction having a second numerator and a second denominator (27), wherein the second numerator includes the dot product of the first channel (101) and the second channel (102) in the subband, and the second denominator includes the dot product.

2. The apparatus according to claim 1, wherein the parameter calculator (140) is configured to calculate the side gain (141) and the residual gain (142) for each subband of a plurality of subbands of the first channel (101) and the second channel (102).

3. The parameter calculator (140) is configured to calculate the side gain (141) using a first fraction (26), wherein the first numerator includes the difference between the first amplitude-related characteristic of the first channel (101) and the second amplitude-related characteristic of the second channel (102), and the first denominator includes the sum of the first amplitude-related characteristic of the first channel (101), the second amplitude-related characteristic of the second channel (102), and the value derived from the dot product, or The parameter calculator (140) is configured to calculate the residual gain (142) using a second fraction having a second numerator and a second denominator (27), wherein the second numerator includes the difference between the weighted sum of the first amplitude-related characteristic of the first channel (101) and the second amplitude-related characteristic of the second channel (102) and the dot product, and the second denominator includes the sum of the amplitude-related characteristic of the first channel (101), the amplitude-related characteristic of the second channel (102), and the value derived from the dot product. The apparatus according to claim 1.

4. The apparatus according to claim 1, wherein the parameter calculator (140) is configured to calculate the side gain (141) with respect to a subband, and (28) use the side gain (141) of the subband to calculate the residual gain (142) with respect to the subband.

5. The apparatus according to claim 1, wherein the parameter calculator (140) is configured to calculate the subband representations of the first channel (101) and the second channel (102) as a sequence of complex-valued spectra, each spectrum in the sequence of complex-valued spectra relating to a time frame of the first channel (101) or the second channel (102), and the time frames relating to the sequence of complex-valued spectra are adjacent and overlapping within the sequence of complex-valued spectra.

6. The apparatus according to claim 1, wherein the parameter calculator (140) is configured to calculate a first amplitude-related feature and a second amplitude-related feature by squaring the magnitudes of the complex spectral values ​​in the subband and summing the multiple squared magnitudes in the subband.

7. The apparatus according to claim 1, wherein the parameter calculator (140) is configured to calculate the dot product between the first channel (101) and the second channel (102) in the subband by summing a plurality of products in the subband, wherein each of the plurality of products includes the spectral value in the frequency bin of the first channel (101) and the complex conjugate spectral value in the frequency bin of the second channel (102), and by forming the magnitude of the result of the sum of the plurality of products.

8. An apparatus for encoding a multichannel signal (100) including at least two channels (101, 102), A downmixer (120) is used to calculate a downmix signal (122) from the multi-channel signal (100), A parameter calculator (140) for calculating side gains (141) from the first channel (101) of the at least two channels (101, 102) and the second channel (102) of the at least two channels (101, 102), and for calculating residual gains (142) from the first channel (101) and the second channel (102), An output interface (160) for generating an output signal, wherein the output signal includes information relating to the downmix signal (122), information relating to the side gain (141), and information relating to the residual gain (142). Equipped with, The side gain (141) represents a side prediction gain applicable to the mid-signals of the first channel (101) and the second channel (102) in order to predict the side signals of the first channel (101) and the second channel (102), and the residual gain (142) represents a residual prediction gain that shows the amplitude-related characteristics of the residual signal of the prediction of the side signal by the mid-signal using the side gain (141). The parameter calculator (140) determines that the value of the side gain (141) is given by the following formula: [Math 1] The side gain (141) is configured to be calculated so that it is within ±20% of the value determined based on, or The parameter calculator (140) determines that the value relating to the residual gain (142) is given by the following formula: [Math 2] The system is configured to calculate the residual gain (142) such that it is within ±20% of the value determined based on the following: t is the frame index, b is the subband index, and E L The energy of the first channel (101) in the frame t and the subband b is E R An apparatus in which X is the energy of the second channel (102) in the frame t and the subband b, and X is the absolute value of the dot product between the first channel (101) and the second channel (102) in the frame t and the subband b.

9. An apparatus for encoding a multichannel signal (100) including at least two channels (101, 102), A downmixer (120) is used to calculate a downmix signal (122) from the multi-channel signal (100), A parameter calculator (140) for calculating side gains (141) from the first channel (101) of the at least two channels (101, 102) and the second channel (102) of the at least two channels (101, 102), and for calculating residual gains (142) from the first channel (101) and the second channel (102), An output interface (160) for generating an output signal, wherein the output signal includes information relating to the downmix signal (122), information relating to the side gain (141), and information relating to the residual gain (142). Equipped with, The side gain (141) represents a side prediction gain applicable to the mid-signals of the first channel (101) and the second channel (102) in order to predict the side signals of the first channel (101) and the second channel (102), and the residual gain (142) represents a residual prediction gain that shows the amplitude-related characteristics of the residual signal of the prediction of the side signal by the mid-signal using the side gain (141). The apparatus is configured such that the downmixer (120) calculates the downmix signal (122) using phase compensation such that only channels with lower energy are rotated, or channels with higher energy are rotated more strongly than the downmix signal (122).

10. The downmixer (120) is configured to calculate the inter-channel phase difference (30), The downmixer (120) is configured to calculate the phase rotation parameter (32), The downmixer (120) is configured to weight the first channel (101) and the second channel (102) (34) using the inter-channel phase difference and the phase rotation parameter when calculating the downmix signal (122). The apparatus according to claim 9.

11. The apparatus according to claim 10, wherein the downmixer (120) is configured to calculate the phase rotation parameter (32) based on the side gain (141) determined by the parameter calculator (140).

12. The apparatus according to claim 10, wherein the downmixer (120) is configured to calculate the inter-channel phase difference for each subband of the frame (30), and the downmixer (120) is configured to calculate the phase rotation parameter for each subband of the frame (32).

13. The downmixer (120) has a phase rotation parameter that is given by the following equation: [Math 3] The system is configured to calculate the phase rotation parameter such that it is within ±20% of the value determined by, where atan2 is the atan2 function, β is the phase rotation parameter, IPD is the inter-channel phase difference, t is the frame index, b is the subband index, and g t,b The apparatus according to claim 9, wherein is the side gain (141) with respect to the frame t and the subband b, and A is a value between 0.1 and 100, or a value between -0.1 and -100.

14. The downmixer (120) and the downmix signal (122) are calculated using the following formula: [Math 4] The downmix signal (122) is calculated (34) such that it has a value within ±20% of the value determined by M t,k The downmix signal (122) is with respect to frame t and frequency bin k, L t,k The first channel (101) is the frame t and the frequency bin k, and R t,k The second channel (102) is the frame t and the frequency bin k, and IPD t,b The apparatus according to claim 9, wherein is the inter-channel phase difference with respect to the frame t and the subband b, which include the frequency bin k, and β is the phase rotation parameter.

15. The apparatus according to claim 1, wherein the output interface (160) comprises a waveform encoder configured to waveform encode the downmix signal (122) in order to obtain the information relating to the downmix signal (122).

16. The parameter calculator (140) is configured to calculate the side gain (141) and the residual gain (142) such that the residual gain (142) depends on the side gain (141), The output interface (160) is configured to quantize the side gain (141) and then the residual gain (142), wherein the quantization step for the residual gain (142) depends on the value of the side gain (141). The apparatus according to claim 1.

17. An apparatus for encoding a multichannel signal (100) including at least two channels (101, 102), A downmixer (120) is used to calculate a downmix signal (122) from the multi-channel signal (100), A parameter calculator (140) for calculating side gains (141) from the first channel (101) of the at least two channels (101, 102) and the second channel (102) of the at least two channels (101, 102), and for calculating residual gains (142) from the first channel (101) and the second channel (102), An output interface (160) for generating an output signal, wherein the output signal includes information relating to the downmix signal (122), information relating to the side gain (141), and information relating to the residual gain (142). Equipped with, The side gain (141) represents a side prediction gain applicable to the mid-signals of the first channel (101) and the second channel (102) in order to predict the side signals of the first channel (101) and the second channel (102), and the residual gain (142) represents a residual prediction gain that shows the amplitude-related characteristics of the residual signal of the prediction of the side signal by the mid-signal using the side gain (141). The parameter calculator (140) is configured to calculate the side gain (141) and the residual gain (142) such that the residual gain (142) depends on the side gain (141), The apparatus wherein the output interface is configured to perform coupled quantization using groups of quantization points, each group of quantization points being defined by a fixed amplitude correlation ratio between the first channel (101) and the second channel (102).

18. The parameter calculator (140) is configured to calculate the side gain (141) such that the side gain (141) has a value range between -1 and +1. The output interface (160) is configured to use a code that has a sign bit and a side gain value that is either positive or negative only. The apparatus according to claim 17.

19. The output interface (160) is The inter-channel level difference between the first channel (101) and the second channel (102) is calculated (50), Identify the group of quantization points that coincide with the inter-channel level difference (52), Search only within the identified group (54), A codeword representing the quantization side gain and the quantization residual gain is obtained by combining the sign bit, the group identification, and the identification of the point within the identified group (56). The apparatus according to claim 17, configured as follows.

20. The apparatus according to claim 17, wherein the codebook used by the output interface (160) comprises a code table having a number of entries, each entry being identified by a binary codeword, each binary codeword having a sign bit, a first group of bits identifying a group of quantization points, and a second group of bits identifying a quantization point within the group of quantization points.

21. The apparatus according to claim 17, wherein the codebook used by the output interface (160) comprises 16 groups of quantization points of 8 quantization points per group, and the codeword of the codebook is an 8-bit codeword having a single sign bit (57a), a 4-bit group (57b) that identifies a group from the 16 groups, and a 3-bit group (57c) that identifies a quantization point within the identified group of quantization points.

22. The parameter calculator (140) The side signals are calculated from the first channel (101) and the second channel (102) (60), Multiple residual gains are determined from the difference between the side signal and the downmix signal (122) weighted by multiple different test side gains (61), The residual signal is selected as the side gain (141) from among the multiple different test side gains that satisfy the defined conditions (62), The residual gain (142) is calculated from a specific residual signal determined using the specific testside gain (63). The apparatus according to claim 1, configured as follows.

23. The apparatus according to claim 22, wherein the residual gain (142) is determined from the energy of the specific residual signal, the energy of the downmix signal (122), or the combined energy of the first channel (101) and the second channel (102).

24. A device for decoding an encoded multichannel signal (200), An input interface (204) for receiving the encoded multichannel signal (200) and obtaining a downmix signal (207), a side gain (206), and a residual gain (205) from the encoded multichannel signal (200), wherein the side gain (206) represents a side prediction gain applicable to the mid-signals of the first and second channels for predicting the side signals of the first and second channels, and the residual gain (205) represents a residual prediction gain showing the amplitude-related characteristics of the residual signal of the prediction of the side signal by the mid-signal using the side gain (206), A residual signal combiner (208) for combining the residual signal (209) using the residual gain (205), An upmixer (212) is used to upmix the downmix signal (207) using the side gain (206) and the residual signal (209) in order to obtain the reconfigured first channel (213) and the reconfigured second channel (214). Equipped with, The residual signal combiner (208) calculates that the value of the residual signal (209) is given by the following formula: [Math 5] The residual signal (209) is calculated such that it is within ±20% of the value determined based on, where res t,k is the residual signal (209) with respect to frame t and frequency bin k. [Math 6] The residual gain (205) is the frame t and the subband b including the frequency bin k, [Number 7] The apparatus wherein the raw signal is the residual signal (209), and g norm is an energy adjustment coefficient which may or may not exist.

25. The residual signal combiner (208) is configured to weight the downmix signal (207) of the preceding frame using the residual gain (205) for the current frame in order to obtain the residual signal (209) for the current frame, or The system is configured to weight (80) an uncorrelated signal derived from (80) the current frame or from one or more preceding frames using the residual gain (205) for the current frame in order to obtain the residual signal (209) for the current frame, The apparatus according to claim 24.

26. The apparatus according to claim 24, wherein the residual signal combiner (208) is configured to calculate the residual signal (209) such that the energy of the residual signal (209) is equal to the signal energy indicated by the residual gain (205).

27. ​​The energy adjustment coefficient g norm The following equation exists: [Number 8] An energy adjustment coefficient having a value within ±20% of the value determined by, [Number 9] This is the energy of the downmix signal (207) with respect to frame t and subband b, [Number 10] This is either the energy of the residual signal (209) with respect to the subband b and the frame t, or The raw signal relating to the residual signal (209) is given by the following equation: [Math 11] It is determined based on, [Number 12] This is the raw signal relating to the residual signal (209), [Number 13] is frame t-t b and said downmix signal (207) relating to frequency bin k, d b is a frame delay greater than 0, The apparatus according to claim 24.

28. A device for decoding an encoded multichannel signal (200), An input interface (204) for receiving the encoded multichannel signal (200) and obtaining a downmix signal (207), a side gain (206), and a residual gain (205) from the encoded multichannel signal (200), wherein the side gain (206) represents a side prediction gain applicable to the mid-signals of the first and second channels for predicting the side signals of the first and second channels, and the residual gain (205) represents a residual prediction gain showing the amplitude-related characteristics of the residual signal of the prediction of the side signal by the mid-signal using the side gain (206), A residual signal combiner (208) for combining the residual signal (209) using the residual gain (205), An upmixer (212) is used to upmix the downmix signal (207) using the side gain (206) and the residual signal (209) in order to obtain the reconfigured first channel (213) and the reconfigured second channel (214). Equipped with, The upmixer (212) uses the following formula to determine whether the reconfigured first channel (213) and the reconfigured second channel (214). [Number 14] [Number 15] The system is configured to calculate the reconstructed first channel (213) and the reconstructed second channel (214) such that they have values ​​within ±20% of the value determined by the system. [Number 16] The downmix signal (207) is the frame t and frequency bin k, [Number 17] The reconstructed first channel (213) is the frame t and the frequency bin k, [Number 18] The reconstructed second channel (214) is the frame t and the frequency bin k, [Number 19] This is the side gain (206) with respect to frame t and subband b, [Number 20] The residual gain (205) with respect to the frame t and the subband b is g norm This is an energy adjustment coefficient that may or may not exist, [Number 21] The apparatus is a raw signal relating to the residual signal (209) with respect to the frame t and the frequency bin k.

29. A device for decoding an encoded multichannel signal (200), An input interface (204) for receiving the encoded multichannel signal (200) and obtaining a downmix signal (207), a side gain (206), and a residual gain (205) from the encoded multichannel signal (200), wherein the side gain (206) represents a side prediction gain applicable to the mid-signals of the first and second channels for predicting the side signals of the first and second channels, and the residual gain (205) represents a residual prediction gain showing the amplitude-related characteristics of the residual signal of the prediction of the side signal by the mid-signal using the side gain (206), A residual signal combiner (208) for combining the residual signal (209) using the residual gain (205), An upmixer (212) is used to upmix the downmix signal (207) using the side gain (206) and the residual signal (209) in order to obtain the reconfigured first channel (213) and the reconfigured second channel (214). Equipped with, The input interface (204) is configured to acquire inter-channel phase difference values ​​from the encoded multi-channel signal (200), The residual signal combiner (208) or the upmixer (212) is configured to apply the inter-channel phase difference value when calculating the residual signal or the reconstructed first channel (213) and the reconstructed second channel (214), The upmixer (212) calculates the phase rotation parameter from the inter-channel phase difference value, In the first mode, the phase rotation parameter is applied when calculating the reconstructed first channel (213), and in the second mode, the inter-channel phase difference value and / or the phase rotation parameter is applied when calculating the reconstructed second channel (214). An apparatus configured such that the first form differs from the second form.

30. The upmixer (212) has the phase rotation parameter as follows: [Number 22] The system is configured to calculate the phase rotation parameter such that it is within ±20% of the value determined by, where atan2 is the atan2 function, β is the phase rotation parameter, IPD is the inter-channel phase difference value, t is the frame index, b is the subband index, and g t,b The apparatus according to claim 29, wherein is the side gain (206) with respect to the frame t and the subband b, and A is a value between 0.1 and 100 or a value between -0.1 and -100.

31. The apparatus according to claim 24, wherein the input interface (204) is configured to extract a codeword, the codeword comprising a combination of a quantized side gain and a quantized residual gain, and the input interface (204) is configured to dequantize the combined codeword using a predefined codebook to obtain the side gain (206) and the residual gain (205) to be used by the residual signal combiner (208) and the upmixer (212).

32. The apparatus according to claim 31, wherein the predefined codebook used by the input interface (204) comprises 16 groups of quantization points of 8 quantization points per group, and the codeword of the predefined codebook is an 8-bit codeword having a single sign bit (57a), a 4-bit group (57b) identifying a group from the 16 groups, and a 3-bit group (57c) identifying a quantization point within the identified group of quantization points.

33. The upmixer (212) is configured to calculate the reconstructed first channel (213) and the reconstructed second channel (214) in the spectral region. The apparatus further comprises a spectrum-to-time converter (216) for converting the reconstructed first channel (213) and the reconstructed second channel (214) into the time domain. The apparatus according to claim 24.

34. The spectrum-time converter (216) For each of the reconstructed first channel (213) and the reconstructed second channel (214), the subsequent frames are converted into a time sequence of time frames (1310), A composite window is used to weight each time frame in the time sequence in order to obtain subsequent windowed time frames (1312), The subsequent windowed time frames are superimposed and added to obtain the first time block of the reconstructed first channel (213) and the second time block of the reconstructed second channel (214) (1314). The apparatus according to claim 33, configured as described above.

35. A method for encoding a multichannel signal (100) including at least two channels (101, 102), The steps include: (120) calculating a downmix signal (122) from the multi-channel signal (100); Step (140) of calculating the side gain (141) from the first channel (101) of the at least two channels (101, 102) and the second channel (102) of the at least two channels (101, 102), and calculating the residual gain (142) from the first channel (101) and the second channel (102), A step (160) of generating an output signal, wherein the output signal includes information relating to the downmix signal (122), information relating to the side gain (141), and information relating to the residual gain (142). Includes, The side gain (141) represents a side prediction gain applicable to the mid-signals of the first channel (101) and the second channel (102) in order to predict the side signals of the first channel (101) and the second channel (102), and the residual gain (142) represents a residual prediction gain that shows the amplitude-related characteristics of the residual signal of the prediction of the side signal by the mid-signal using the side gain (141). The step of calculating the side gain includes a step (26) of calculating the side gain (141) using a first fraction having a first numerator and a first denominator, wherein the first numerator includes a first amplitude-related characteristic of the first channel (101) and a second amplitude-related characteristic of the second channel (102), and the first denominator includes the first amplitude-related characteristic of the first channel (101), the second amplitude-related characteristic of the second channel (102), and the dot product of the first channel (101) and the second channel (102) in the subband, or The step of calculating the residual gain includes a step (27) of calculating the residual gain (142) using a second fraction having a second numerator and a second denominator, wherein the second numerator includes the dot product of the first channel (101) and the second channel (102) in the subband, and the second denominator includes the dot product, or The step of calculating the side gain is that the value relating to the side gain (141) is given by the following formula, [Number 23] The process includes the step of calculating the side gain (141) such that it is within ±20% of the value determined based on, The step of calculating the residual gain is that the value relating to the residual gain (142) is given by the following formula: [Number 24] The step includes calculating the residual gain (142) such that it is within ±20% of the value determined based on, t is the frame index, b is the subband index, EL is the energy of the first channel (101) in frame t and subband b, ER is the energy of the second channel (102) in frame t and subband b, and X is the absolute value of the inner product between the first channel (101) and the second channel (102) in frame t and subband b, or The step of calculating the downmix signal includes a step of calculating the downmix signal (122) using phase compensation such that when calculating the downmix signal (122), only channels with lower energy are rotated or rotated more strongly than channels with higher energy, or A method wherein the steps of calculating the side gain and the steps of calculating the residual gain include the step of calculating the side gain (141) and the residual gain (142) such that the residual gain (142) depends on the side gain (141), and the steps of generating include the step of performing coupled quantization using a group of quantization points, where each group of quantization points is defined by a fixed amplitude correlation ratio between the first channel (101) and the second channel (102).

36. A method for decoding an encoded multichannel signal (200), Step (204) of receiving the encoded multichannel signal (200) and obtaining a downmix signal (207), a side gain (206), and a residual gain (205) from the encoded multichannel signal (200), wherein the side gain (206) represents a side prediction gain applicable to the mid-signals of the first and second channels in order to predict the side signals of the first and second channels, and the residual gain (205) represents a residual prediction gain that shows the amplitude-related characteristics of the residual signal of the prediction of the side signal by the mid-signal using the side gain (206); The steps include: (208) synthesizing the residual signal (209) using the residual gain (205); Step (212) upmixes the downmix signal (207) using the side gain (206) and the residual signal (209) in order to obtain the reconstructed first channel (213) and the reconstructed second channel (214). Includes, The step of synthesis is such that the value of the residual signal (209) is given by the following formula: [Number 25] The step includes calculating the residual signal (209) such that it is within ±20% of the value determined based on, where res t,k is the residual signal (209) with respect to frame t and frequency bin k. [Number 26] The residual gain (205) is the frame t and the subband b including the frequency bin k, [Number 27] This is the raw signal relating to the residual signal (209), and g norm is an energy adjustment coefficient which may or may not exist, or The upmixing step is such that the reconfigured first channel (213) and the reconfigured second channel (214) are given by the following formula: [Number 28] [Number 29] The steps include calculating the reconstructed first channel (213) and the reconstructed second channel (214) such that they have values ​​within ±20% of the value determined by [Number 30] The downmix signal (207) is the frame t and frequency bin k, [Number 31] The reconstructed first channel (213) is the frame t and the frequency bin k, [Number 32] The reconstructed second channel (214) is the frame t and the frequency bin k, [Number 33] This is the side gain (206) with respect to frame t and subband b, [Number 34] This is the residual gain (205) with respect to the frame t and the subband b, and g norm is an energy adjustment coefficient which may or may not exist. [Number 35] is the raw signal relating to the residual signal (209) with respect to the frame t and the frequency bin k, or The receiving step includes the step of obtaining an inter-channel phase difference value from the encoded multi-channel signal (200), A method wherein the step of synthesizing or the step of upmixing includes a step of applying the inter-channel phase difference value when calculating the residual signal or the reconstructed first channel (213) and the reconstructed second channel (214), and the step of upmixing includes a step of calculating a phase rotation parameter from the inter-channel phase difference value, applying the phase rotation parameter when calculating the reconstructed first channel (213) in a first manner, and applying the inter-channel phase difference value and / or the phase rotation parameter when calculating the reconstructed second channel (214) in a second manner, wherein the first manner is different from the second manner.

37. A computer program for performing the method according to claim 35 or the method according to claim 36, when running on a computer or processor.

Images