Audio Encoder for Encoding a Multi-Channel Signal and Audio Decoder for Decoding the Encoded Audio Signal - Patent application

Abstract

An improved audio encoder and decoder, encoding method and decoding method, and program for encoding a multi-channel signal are provided. The encoder includes a downmixer 12 for downmixing a multi-channel signal 4 to obtain a downmix signal 14, a linear prediction domain core encoder 16 for encoding the downmix signal 14, a filter bank 82 for generating a spectral representation of the multi-channel signal 4, and a joint multi-channel encoder 18 for processing the spectral representation including a low band and a high band of the multi-channel signal to generate multi-channel information 20. The downmix signal 14 has a low band and a high band, and the linear prediction domain core encoder 16 applies a bandwidth extension process to parametrically encode the high band.

Description

[Technical field]

[0001] The present invention relates to an audio encoder for encoding a multi-channel audio signal and an audio decoder for decoding the encoded audio signal.Embodiments relate to multi-channel encoding in an LPD using a filter bank for multi-channel processing that is not used for bandwidth extension. [Background technology]

[0002] Perceptual coding of audio signals for the purpose of data reduction for efficient storage or transmission of these signals is a widely used practice. In particular, when the highest efficiency needs to be achieved, coders that closely adapt to the signal input characteristics are used. One example is the MPEG-D USAC core coder, which can be configured to primarily use ACELP (Algebraic Code-Excited Linear Prediction) coding for speech signals, TCX (Transform Coded Excitation) for background noise and mix signals, and AAC (Advanced Audio Coding) for music content. All three internal coder configurations are switched on the fly in a signal-adaptive manner corresponding to the signal content.

[0003] Furthermore, joint multi-channel coding techniques (such as mid / side coding) or, for the most efficient, parametric coding techniques are used. Parametric coding techniques essentially aim at recreating a perceptually equivalent audio signal rather than a faithful reconstruction of a given waveform. Examples include noise filling, bandwidth expansion and spatial audio coding.

[0004] When combining a signal-adaptive core coder with any one of the joint multi-channel coding techniques or parametric coding techniques of state-of-the-art coders, the core coder is switched to match the signal characteristics, but the choice of the multi-channel coding technique, such as M / S-stereo, spatial audio coding or parametric stereo, remains fixed and independent of the signal characteristics. These techniques are usually used in the core coder and in the core encoder as pre-processor and in the core decoder as post-processor (both without knowing the actual choice of the core coder).

[0005] On the other hand, the choice of parametric coding technique for bandwidth extension sometimes depends on the signal. For example, techniques applied in the time domain are more efficient for speech signals, while frequency domain processing is more relevant for other signals. In such cases, the adopted multi-channel coding technique must be compatible with both types of bandwidth extension techniques.

[0006] State-of-the-art related topics include: For the MPEG-D USAC core encoder, PS and MPS are used as pre- and post-processors. MPEG-D USAC standard MPEG-H 3D Audio Standard [Prior art documents] [Non-patent literature]

[0007] [Non-Patent Document 1] ISO / IEC DIS23003-3, Usac [Non-Patent Document 2] ISO / IEC DIS23008-3, 3D Audio Summary of the Invention [Problem to be solved by the invention]

[0008] In MPEG-D USAC, a switchable core coder is described. However, in USAC, the multi-channel coding technique is defined as a fixed choice common to the entire core coder, independent of its internal switch of coding principle, be it ACELP or TCX ("LPD") or AAC ("FD"). Thus, if a switched core coder configuration is required, the coder is restricted to use parametric multi-channel coding (PS) for the entire signal until the end. However, for coding of e.g. music signals, it is more appropriate to use joint stereo coding, which can rather dynamically switch between L / R (left / right) and M / S (middle / side) schemes per frequency band and per frame.

[0009] Thus, there is a need for improved approaches.

[0010] It is an object of the present invention to provide an improved concept for processing audio signals. This object is solved by the subject matter of the independent claims. [Means for solving the problem]

[0011] The invention is based on the discovery that a (time domain) parametric encoder using a multi-channel coder is advantageous for parametric multi-channel audio coding. The multi-channel coder may be a multi-channel residual coder that reduces the bandwidth for the transmission of coding parameters compared to individual coding per channel. For example, this is advantageously used in combination with a frequency domain joint multi-channel audio coder. The time domain joint multi-channel coding technique and the frequency domain joint multi-channel coding technique are combined, so that for example a frame-based decision can lead to a time-based or a frequency-based coding period for the current frame. That is to say, the embodiment shows an improved concept for combining a switchable core coder using joint multi-channel coding and parametric spatial audio coding into a fully switchable perceptual coder, which allows to use different multi-channel coding techniques in dependence of the core coder selection. This is advantageous because, in contrast to existing methods, the embodiment shows a multi-channel coding technique that is switched simultaneously immediately to the core coder, and therefore closely matches and adapts to the selection of the core coder. Thus, the described problems that appear due to a fixed selection of the multi-channel coding technique are avoided. Furthermore, a fully switchable combination of a given core coder and associated adapted multi-channel coding techniques is possible. A coder such as AAC (Advanced Audio Coding), using e.g. L / R or M / S stereo coding, may code a music signal in a dedicated joint stereo or frequency domain (FD) core coder using multi-channel coding, e.g. M / S stereo. This decision is applied separately for each frequency band within each audio frame. In the case of e.g. a speech signal, the core coder immediately switches to a linear predictive decoding (LPD) core coder and its associated different, e.g. parametric stereo coding technique.

[0012] The embodiment shows only stereo processing in the mono LPD path, and a stereo signal based seamless switching scheme that combines the output of the stereo FD path with the output from the LPD core coder and its dedicated stereo encoding, which is advantageous since it allows for seamless coder switching without artifacts.

[0013] An embodiment relates to an encoder for encoding a multi-channel signal. The encoder includes a linear prediction domain encoder and a frequency domain encoder. The encoder further includes a controller for switching between the linear prediction domain encoder and the frequency domain encoder. The linear prediction domain encoder further includes a downmixer for downmixing the multi-channel signal to obtain a downmix signal, a linear prediction domain core encoder for encoding the downmix signal, and a first multi-channel encoder for generating first multi-channel information from the multi-channel signal. The frequency domain encoder includes a second combined multi-channel encoder for encoding second multi-channel information from the multi-channel signal. The second multi-channel encoder is different from the first multi-channel encoder. The controller is configured such that the portions of the multi-channel signal are represented either by the encoded frames of the linear prediction domain encoder or by the encoded frames of the frequency domain encoder. The linear prediction domain encoder includes an ACELP core encoder and, for example, a parametric stereo encoding algorithm as the first combined multi-channel encoder. The frequency domain encoder includes, for example, an AAC core encoder using, for example, L / R or M / S processing as the second combined multi-channel encoder. The controller analyzes the multi-channel signal for frame characteristics, e.g., speech or music, and determines which individual frame, a series of frames, or a portion of the multi-channel audio signal, either a linear predictive domain encoder or a frequency domain encoder is used to encode this portion of the multi-channel audio signal.

[0014] The embodiment further illustrates an audio decoder for decoding an encoded audio signal. The audio decoder includes a linear prediction domain decoder and a frequency domain decoder. The audio decoder further includes a first combined multi-channel decoder for generating a first multi-channel representation using an output of the linear prediction domain decoder and the multi-channel information, and a second multi-channel decoder for generating a second multi-channel representation using an output of the frequency domain decoder and the second multi-channel information. The audio decoder further includes a first combiner for combining the first multi-channel representation and the second multi-channel representation to obtain a decoded audio signal. The combiner performs seamless and artifact-free switching between the first multi-channel representation, e.g. a linear prediction multi-channel audio signal, and the second multi-channel representation, e.g. a frequency domain decoded multi-channel audio signal.

[0015] An embodiment shows a combination of ACELP / TCX coding in the LPD path with a dedicated stereo coding and an independent AAC stereo coding in the frequency domain path in a switchable audio coder. Furthermore, an embodiment shows a seamless instantaneous switch between LPD and FD stereo. Another embodiment relates to an independent selection of joint multi-channel coding for different signal content types. For example, for speech that is mainly coded using the LPD path, parametric stereo is used, whereas for music that is coded in the FD path, a more adaptive stereo coding is used. It can dynamically switch between L / R and M / S schemes on a frequency band-by-frequency band and frame-by-frame basis.

[0016] According to an embodiment, simple parametric stereo is adequate for speech, which is primarily coded using the LPD path and is always located in the center of the stereo image, whereas music, which is coded in the FD path, always has a more refined spatial distribution and can benefit from a more adaptive stereo coding, which can dynamically switch between L / R and M / S schemes on a frequency band-by-frequency band and frame-by-frame basis.

[0017] Another embodiment shows an audio encoder including a downmixer (12) for downmixing a multi-channel signal to obtain a downmix signal, a linear prediction domain core encoder for encoding the downmix signal, a filter bank for generating a spectral representation of the multi-channel signal, and a joint multi-channel encoder for generating multi-channel information from the multi-channel signal. The downmix signal has a low band and a high band. The linear prediction domain core encoder is configured to apply a bandwidth extension process to parametrically encode the high band. Furthermore, the multi-channel encoder is configured to process a spectral representation including the low band and the high band of the multi-channel signal. This is advantageous because each parametric encoding can use its optimal time-frequency decomposition for obtaining its parameters. This is implemented, for example, using a combination of ACELP (Algebraic Code Excited Linear Prediction) + TDBWE (Time Domain Bandwidth Extension). ACELP encodes the low band of the audio signal, TDBWE encodes the high band of the audio signal, and parametric multi-channel encoding with an external filter bank (e.g. DFT). This combination is particularly efficient since it is known that the best bandwidth expansion for speech is in the time domain and the multi-channel processing should be in the frequency domain. Since ACELP+TDBWE does not have any time-to-frequency converters, external filter banks or transformations such as DFT are advantageous. Furthermore, the framing of the multi-channel processor is the same as that used in ACELP. Even if the multi-channel processing is done in the frequency domain, the time resolution for the computation of its parameters or downmix is ​​ideally close to or even equal to the framing of ACELP.

[0018] The described embodiment is advantageous because independent selection of joint multi-channel coding is applied for different signal content types.

[0019] Embodiments of the present invention will now be described with reference to the accompanying drawings. [Brief description of the drawings]

[0020] [Figure 1] FIG. 1 shows a schematic block diagram of an encoder for encoding a multi-channel audio signal. [Diagram 2] FIG. 2 shows a schematic block diagram of a linear prediction domain encoder according to an embodiment. [Diagram 3] FIG. 3 shows a schematic block diagram of a frequency domain encoder according to an embodiment. [Figure 4] FIG. 4 shows a schematic block diagram of an audio encoder according to an embodiment. [Figure 5a] FIG. 5a shows a schematic block diagram of an active downmixer according to an embodiment. [Figure 5b] FIG. 5b shows a schematic block diagram of a passive downmixer according to an embodiment. [Figure 6] FIG. 6 shows a schematic block diagram of a decoder for decoding an encoded audio signal. [Figure 7] FIG. 7 shows a schematic block diagram of a decoder according to an embodiment. [Figure 8] FIG. 8 shows a schematic block diagram of a method for encoding a multi-channel signal. [Figure 9] FIG. 9 shows a schematic block diagram of a method for decoding an encoded audio signal. [Figure 10] FIG. 10 shows a schematic block diagram of an encoder for encoding a multi-channel signal according to another embodiment. [Figure 11] FIG. 11 shows a schematic block diagram of a decoder for decoding an encoded audio signal according to another embodiment. [Figure 12] FIG. 12 shows a schematic block diagram of an audio coding method for coding a multi-channel signal according to another embodiment. [Figure 13]FIG. 13 shows a schematic block diagram of a method for decoding an encoded audio signal according to another embodiment. [Figure 14] Figure 14 shows an overview timing diagram of the seamless switchover from frequency domain coding to LPD coding. [Figure 15] Figure 15 shows an overview timing diagram of the seamless switchover from frequency domain decoding to LPD domain decoding. [Figure 16] Figure 16 shows an overview timing diagram of the seamless switchover from LPD coding to frequency domain coding. [Figure 17] Figure 17 shows an overview timing diagram of the seamless switchover from LPD decoding to frequency domain decoding. [Figure 18] FIG. 18 shows a schematic block diagram of an encoder for encoding a multi-channel signal according to another embodiment. [Figure 19] FIG. 19 shows a schematic block diagram of a decoder for decoding an encoded audio signal according to another embodiment. [Figure 20] FIG. 20 shows a schematic block diagram of an audio coding method for encoding a multi-channel signal according to another embodiment. [Figure 21] FIG. 21 shows a schematic block diagram of a method for decoding an encoded audio signal according to another embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0021] In the following, embodiments of the present invention will be described in more detail. Elements designated with individual numerals having the same or similar functions are associated with the same reference sign.

[0022] FIG. 1 shows a schematic block diagram of an audio encoder 2 for encoding a multi-channel audio signal 4. The audio encoder comprises a linear prediction domain encoder 6, a frequency domain encoder 8 and a controller 10 for switching between the linear prediction domain encoder 6 and the frequency domain encoder 8. The controller analyses the multi-channel signal and decides whether linear prediction domain encoding or frequency domain encoding is advantageous for a portion of the multi-channel signal. That is, the controller is configured such that a portion of the multi-channel signal is represented either by an encoded frame of the linear prediction domain encoder or by an encoded frame of the frequency domain encoder. The linear prediction domain encoder comprises a downmixer 12 for downmixing the multi-channel signal 4 to obtain a downmix signal 14. The linear prediction domain encoder further comprises a linear prediction domain core encoder 16 for encoding the downmix signal. Furthermore, the linear prediction domain encoder comprises a first combined multi-channel encoder 18 for generating first multi-channel information 20 from the multi-channel signal 4, for example comprising an ILD (Inter-ear Level Difference) parameter and / or an IPD (Inter-ear Phase Difference) parameter. The multi-channel signal is, for example, a stereo signal. The downmixer converts the stereo signal to a mono signal. The linear prediction domain core encoder encodes the mono signal. The first joint multi-channel encoder generates stereo information for the encoded mono signal as first multi-channel information. The frequency domain encoder and controller are optional when compared with the alternative embodiment described with respect to Figures 10 and 11. However, it is advantageous to use a frequency domain encoder and controller for signal adaptive switching between time domain and frequency domain coding.

[0023] Furthermore, the frequency domain encoder 8 includes a second combined multi-channel encoder 22 for generating second multi-channel information 24 from the multi-channel signal 4. The second combined multi-channel encoder 22 is different from the first multi-channel encoder 18. However, the second combined multi-channel encoder 22 obtains second multi-channel information which allows for a second reconstruction quality, higher than the first reconstruction quality of the first multi-channel information obtained by the first multi-channel encoder, for signals better encoded by the second encoder.

[0024] That is, according to an embodiment, the first combined multi-channel encoder 18 is configured to generate a first multi-channel information 20 allowing a first reconstruction quality. The second combined multi-channel encoder 22 is configured to generate a second multi-channel information 24 allowing a second reconstruction quality. The second reconstruction quality is higher than the first reconstruction quality. This is at least relevant for a signal, e.g. a speech signal, which is better encoded by the second multi-channel encoder.

[0025] Thus, the first multi-channel encoder is a parametric combined multi-channel encoder, including, for example, a stereo predictive coder, a parametric stereo encoder, or a rotation-based parametric stereo encoder. Furthermore, the second combined multi-channel encoder is a waveform preserving encoder, such as a band-selective switch for a mid / side or left / right stereo coder. As described in FIG. 1, the encoded downmix signal 26 is sent to an audio decoder and optionally provided to the first combined multi-channel processor. For example, the encoded downmix signal is decoded to obtain a decoded signal, and residual signals from the multi-channel signal before and after the encoding are calculated at the decoder side to enhance the decoded quality of the encoded audio signal. Furthermore, the controller 10 uses the control signals 28a, 28b to control the linear predictive domain encoder and the frequency domain encoder, respectively, after determining a suitable encoding scheme for the current portion of the multi-channel signal.

[0026] Fig. 2 shows a block diagram of a linear prediction domain encoder 6 according to an embodiment. The input to the linear prediction domain encoder 6 is the downmix signal 14 downmixed by the downmixer 12. Furthermore, the linear prediction domain encoder includes an ACELP processor 30 and a TCX processor 32. The ACELP processor 30 is configured to operate on a downsampled downmix signal 34 downsampled by a downsampler 35. Furthermore, a time domain bandwidth extension processor 36 parametrically encodes bands of portions of the downmix signal 14 that are removed from the downsampled downmix signal 34 input into the ACELP processor 30. The time domain bandwidth extension processor 36 outputs parametrically encoded bands 38 of portions of the downmix signal 14. That is, the time domain bandwidth extension processor 36 calculates a parametric representation of frequency bands of the downmix signal 14 that include higher frequencies compared to the cutoff frequency of the downsampler 35. Therefore, the downsampler 35 has other properties to provide the time domain bandwidth extension processor 36 with those frequency bands above the cutoff frequency of the downsampler, or to provide a cutoff frequency to the time domain bandwidth extension (TD-BWE) processor 36 to enable the TD-BWE processor 36 to calculate the parameters 38 for the correct part of the downmix signal 14.

[0027] Furthermore, the TCX processor is configured to operate on a downmix signal that is not downsampled or that is downsampled to a lesser extent than the downsampling for the ACELP processor. Downsampling to a lesser extent than the downsampling for the ACELP processor is downsampling using a higher cutoff frequency. A number of bands of the downmix signal are provided to the TCX processor when compared with the downsampled downmix signal 35 input to the ACELP processor 30. The TCX processor further comprises a first time-to-frequency converter 40, for example an MDCT, DFT or DCT. The TCX processor 32 further comprises a first parameter generator 42 and a first quantizer encoder 44. The first parameter generator 42, for example using an intelligent gap filling (IGF) algorithm, calculates a first parametric representation of a first set of bands 46. The first quantizer encoder 44, for example using a TCX algorithm, calculates a first set of quantized and encoded spectral lines 48 for a second set of bands. That is, the first quantizer encoder parametrically encodes relevant bands, for example tone bands, of the inbound signal. The first parameter generator applies, for example, an IGF algorithm to the remaining bands of the inbound signal to further reduce the bandwidth of the encoded audio signal.

[0028] The linear prediction domain encoder 6 further comprises a linear prediction domain decoder 50 for decoding the downmix signal 14, e.g. represented by an ACELP processed downsampled downmix signal 52 and / or a first parametric representation of the first band set 46 and / or a first set of quantized coded spectral lines 48 for the second band set. The output of the linear prediction domain decoder 50 is an encoded and decoded downmix signal 54. This signal 54 is input to a multi-channel residual coder 56, which uses the encoded and decoded downmix signal 54 to calculate and code a multi-channel residual signal 58. The encoded multi-channel residual signal represents the error between the decoded multi-channel representation using the first multi-channel information and the multi-channel signal before the downmix. The multi-channel residual coder 56 therefore comprises a joint encoder-side multi-channel decoder 60 and a difference processor 62. The joint encoder-side multi-channel decoder 60 uses the first multi-channel information 20 and the encoded and decoded downmix signal 54 to generate a decoded multi-channel signal. The difference processor forms the difference between the decoded multi-channel signal 64 and the multi-channel signal 4 before downmix to obtain a multi-channel residual signal 58. That is to say, a joint encoder-side multi-channel decoder in the audio encoder performs a decoding operation, which is advantageously the same decoding operation performed at the decoder side. Thus, the first joint multi-channel information derived by the audio decoder after transmission is used in the joint encoder-side multi-channel decoder for decoding the coded downmix signal. The difference processor 62 calculates the difference between the decoded joint multi-channel signal and the original multi-channel signal 4. The coded multi-channel residual signal 58 enhances the decoding quality of the audio decoder, since the difference between the decoded signal and the original signal, for example for parametric coding, is reduced by the knowledge of the difference between these two signals.This enables the first combined multi-channel encoder to operate in such a way that multi-channel information for the entire bandwidth of the multi-channel audio signal is derived.

[0029] Furthermore, the downmix signal 14 comprises a low band and a high band. The linear prediction domain encoder 6 is configured to apply a bandwidth extension process, for example using a time domain bandwidth extension processor 36 for parametrically encoding the high band. The linear prediction domain decoder 6 is configured to obtain only a low band signal representing the low band of the downmix signal 14 as the encoded and decoded downmix signal 54. The encoded multi-channel residual signal only has frequencies within the low band of the multi-channel signal before the downmix. That is to say, the bandwidth extension processor calculates a bandwidth extension parameter for the frequency band higher than the cutoff frequency. The ACELP processor encodes the frequencies below the cutoff frequency. The decoder is therefore configured to reconstruct the higher frequencies based on the encoded low band signal and the bandwidth parameter 38.

[0030] According to another embodiment, the multi-channel residual coder 56 calculates a side signal. The downmix signal is a corresponding intermediate signal of the M / S multi-channel audio signal. The multi-channel residual coder therefore calculates and codes the difference between the calculated side signal, calculated from the full-band spectral representation of the multi-channel audio signal obtained by the filter bank 82, and a predicted side signal of a multiple of the coded and decoded downmix signal 54. The multiple represented by the prediction information becomes part of the multi-channel information. However, the downmix signal only contains the low-band signal. The residual coder therefore further calculates a residual (or side) signal for the high-band. This is performed, for example, by simulating a time-domain bandwidth expansion, as is done in a linear prediction domain core encoder. Alternatively, it is performed by predicting the side signal as the difference between the calculated (full-band) side signal and the calculated (full-band) intermediate signal. The prediction factor is configured to minimize the difference between both signals.

[0031] Fig. 3 shows a schematic block diagram of a frequency domain encoder 8 according to an embodiment. The frequency domain encoder comprises a second time-to-frequency converter 66, a second parameter generator 68 and a second quantizer encoder 70. The second time-to-frequency converter 66 converts the first channel 4a and the second channel 4b of the multi-channel signal into spectral representations 72a, 72b. The first channel spectral representation 72a and the second channel spectral representation 72b are analyzed and divided into a first band set 74 and a second band set 76, respectively. The second parameter generator 68 thus generates a second parametric representation 78 of the second band set 76. The second quantizer encoder generates a quantized and coded representation 80 of the first band set 74. The frequency domain encoder, more precisely the second time-to-frequency converter 66, for example performs an MDCT operation on the first channel 4a and the second channel 4b. The second parameter generator 68 performs an intelligent gap filling algorithm and the second quantizer encoder 70 performs, for example, an AAC operation. Thus, as already described for the linear prediction domain encoder, the frequency domain encoder can be operated in such a way that multi-channel information for the entire bandwidth of the multi-channel audio signal is derived.

[0032] Figure 4 shows a schematic block diagram of an audio encoder 2 according to a preferred embodiment. The LPD path 16 consists of a joint stereo or multi-channel encoding including an "active or passive DMX" downmix calculation 12, indicating that the LPD downmix is ​​active ("frequency selective") or passive ("constant mix factor"), as described in Figure 5. The downmix is ​​further encoded by a switchable mono ACELP / TCX core, supported by either the TD-BWE or IGF module. Note that ACELP operates on the downsampled input audio data 34. Any ACELP initialization due to switching is performed on the downsampled TCX / IGF output.

[0033] Since ACELP does not include any internal time-frequency decomposition, LPD stereo coding adds extra complex modular filter banks by means of an analysis filter bank 82 before LP coding and a synthesis filter bank after LPD decoding. In the preferred embodiment, an oversampled DFT with low overlap area is employed. However, in alternative embodiments, an oversampled time-frequency decomposition with similar temporal resolution can be used. The stereo parameters are then calculated in the frequency domain.

[0034] The parametric stereo coding is performed by the "LPD stereo parameter coding" block 18 which outputs LPD stereo parameters 20 to the bitstream. Optionally, the following block "LPD stereo residual coding" adds a vector quantized low-pass downmix residual 58 to the bitstream.

[0035] The FD path 8 is configured to have joint stereo or multi-channel coding within itself. For joint stereo coding, it reuses its own critically sampled real-valued filter bank 66, i.e. for example MDCT.

[0036] The signals provided to the decoder are for example multiplexed into a single bitstream comprising the encoded downmix signal 26 further comprising at least one of the parametrically encoded time-domain bandwidth extended bands 38, the ACELP processed downsampled downmix signal 52, the first multi-channel information 20, the encoded multi-channel residual signal 58, the first parametric representation of the first set of bands 46, the first set of quantized encoded spectral lines for the second set of bands 48, and the second multi-channel information 24 comprising the quantized encoded representation of the first set of bands 80 and the second parametric representation of the first set of bands 78.

[0037] The embodiment shows an improved method for combining a switchable core encoder, joint multi-channel encoding and parametric spatial audio encoding into a fully switchable perceptual encoder that allows to use different multi-channel encoding techniques depending on the choice of the core encoder. In particular, within the switchable audio encoder, native frequency domain stereo encoding is combined with ACELP / TCX based linear predictive coding with its own dedicated independent parametric stereo encoding.

[0038] 5a and 5b respectively show active and passive downmixers according to an embodiment. The active downmixer operates in the frequency domain, for example using a time-frequency converter 82 to convert the time domain signal 4 into a frequency domain signal. After the downmix, a frequency-to-time transformation (e.g. IDFT) converts the downmixed signal from the frequency domain into a downmix signal 14 in the time domain.

[0039] 5b shows a passive downmixer 12 according to an embodiment. The passive downmixer 12 comprises a summer in which the first channel 4a and the second channel 4b are combined after being weighted with weights 84a and 84b, respectively. Furthermore, the first channel 4a and the second channel 4b are input to a time-to-frequency converter 82 before transmission to the LPD stereo parametric coding.

[0040] That is, the downmixer is configured to convert the multi-channel signal into a spectral representation, the downmixing is performed using the spectral representation or using the time domain representation, and the first multi-channel encoder is configured to use the spectral representation to generate separate first multi-channel information for each band of the spectral representation.

[0041] Fig. 6 shows a schematic block diagram of an audio decoder 102 for decoding an encoded audio signal 103 according to an embodiment. The audio decoder 102 comprises a linear prediction domain decoder 104, a frequency domain decoder 106, a first combined multi-channel decoder 108, a second combined multi-channel decoder 110 and a first combiner 112. The encoded audio signal 103, which is a multiplexed bitstream of the previously described encoder parts, e.g. a frame of an audio signal, is decoded by the combined multi-channel decoder 108 using the first multi-channel information 20 or by the frequency domain decoder 106 and the second combined multi-channel decoder 110 using the second multi-channel information 24. The first combined multi-channel decoder outputs a first multi-channel representation 114 and the output of the second combined multi-channel decoder 110 is the second multi-channel representation 116.

[0042] That is, the first combined multi-channel decoder 108 uses the output of the linear prediction domain encoder and the first multi-channel information 20 to generate a first multi-channel representation 114. The second multi-channel decoder 110 uses the output of the frequency domain decoder and the second multi-channel information 24 to generate a second multi-channel representation 116. Furthermore, the first combiner combines the first multi-channel representation 114 and the second multi-channel representation 116, for example on a frame basis, to obtain a decoded audio signal 118. Furthermore, the first combined multi-channel decoder 108 is, for example, a parametric combined multi-channel decoder using complex prediction, a parametric stereo operation or a rotation operation. The second combined multi-channel decoder 110 is, for example, a waveform-preserving combined multi-channel decoder using a band-selective switch for a mid / side or left / right stereo decoding algorithm.

[0043] Fig. 7 shows a schematic block diagram of a decoder 102 according to another embodiment. Here, the linear prediction domain decoder 102 includes an ACELP decoder 120, a low-band synthesizer 122, an upsampler 124, a time domain bandwidth extension processor 126, or a second combiner 128 for combining the upsampled signal and the bandwidth extension signal. Furthermore, the linear prediction domain decoder includes a TCX decoder 132 and an intelligent gap filling processor 132, which are depicted as one block in Fig. 7. Furthermore, the linear prediction domain decoder 102 includes a full-band synthesizer processor 134 for combining the outputs of the second combiner 128, the TCX decoder 130 and the IGF processor 132. As already shown for the encoder, the time domain bandwidth extension processor 126, the ACELP decoder 120 and the TCX decoder 130 work in parallel to decode the individual transmitted audio information.

[0044] The crosspath 136 is provided to initialize the lowband synthesizer using information derived from the lowband spectrum-time conversion, for example using the frequency-to-time converter 138 from the TCX decoder 130 and the IGF processor 132. By referencing a model of the vocal spaciousness, the ACELP data creates a template of the vocal spaciousness. The TCX data creates a template of the excitation of the vocal spaciousness. The crosspath 136, represented by a lowband frequency-to-time converter, for example an IMDCT decoder, allows the lowband synthesizer 122 to use the shape of the vocal spaciousness and to recalculate or decode the lowband signal with the current excitation encoded. Furthermore, the synthesized lowband is upsampled by the upsampler 124 and combined with the time domain bandwidth extension highband 140, for example using the second combiner 128, to recreate the upsampled frequency, for example to recover the energy for each upsampled band.

[0045] The full-band synthesizer 134 uses the full-band signal of the second combiner 128 and the excitation from the TCX processor 130 to form a decoded downmix signal 142. The first combined multi-channel decoder 108 includes a time-to-frequency converter 144 for converting the output of the linear prediction domain decoder, e.g. the decoded downmix signal 142, into a spectral representation 145. Furthermore, an upmixer, for example implemented in a stereo decoder 146, is controlled by the first multi-channel information 20 to upmix the spectral representation into a multi-channel signal. Furthermore, a frequency-to-time converter 148 converts the result of the upmix into a time representation 114. The time-to-frequency and / or frequency-to-time converter includes, for example, complex or oversampled operations such as DFT or IDFT.

[0046] Furthermore, the first combined multi-channel decoder, or more specifically, the stereo decoder 146, uses the multi-channel residual signal 58, for example provided by the multi-channel encoded audio signal 103, to generate a first multi-channel representation. Furthermore, the multi-channel residual signal comprises a lower bandwidth than the first multi-channel representation. The first combined multi-channel decoder is configured to reconstruct an intermediate first multi-channel representation using the first multi-channel information and to add the multi-channel residual signal to the intermediate first multi-channel representation. That is to say, the stereo decoder 146 comprises a multi-channel decoding using the first multi-channel information 20 after the spectral representation of the decoded downmix signal has been upmixed into the multi-channel signal, and optionally a refinement of the reconstructed multi-channel signal by adding the multi-channel residual signal to the reconstructed multi-channel signal. Thus, the first multi-channel information and the residual signal already act on the multi-channel signal.

[0047] The second combined multi-channel decoder 110 uses as input the spectral representation obtained by the frequency domain decoder. The spectral representation comprises at least the first channel signal 150a and the second channel signal 150b for a number of bands. Furthermore, the second combined multi-channel processor 110 adapts to the number of bands of the first channel signal 150a and the second channel signal 150b. A combined multi-channel operation, for example a mask, represents a left / right or a middle / side combined multi-channel encoding for each band. The combined multi-channel operation is a middle / side or a left / right conversion operation for converting the bands represented by the mask from a middle / side representation to a left / right representation. It performs a conversion of the result of the combined multi-channel operation to a time representation to obtain a second multi-channel representation. Furthermore, the frequency domain decoder comprises a frequency-to-time converter 152, for example an IMDCT operation or in particular a sampled operation. That is, the mask comprises a flag representing for example an L / R or M / S stereo encoding. The second combined multi-channel encoder applies a corresponding stereo encoding algorithm to each audio frame. Optionally, intelligent gap filling is applied to the encoded audio signal in order to further reduce the bandwidth of the encoded audio signal. Thus, for example, the tonal frequency band is encoded in high resolution using the aforementioned stereo encoding algorithm, while other frequency bands are encoded parametrically, for example by using the IGF algorithm.

[0048] That is, in the LPD path 104, the transmitted mono signal is reconstructed by a switchable ACELP / TCX 120 / 130 decoder, for example supported by the TD-BWE 126 or IGF module 132. Any ACELP initialization by switching is performed on the downsampled TCX / IGF output. The output of the ACELP is upsampled to the full sampling rate, for example using the upsampler 124. All signals are mixed in the time domain at a higher sampling rate, for example using the mixer 128, and further processed by the LPD stereo decoder 146 to provide LPD stereo.

[0049] The LPD "stereo decoding" consists of an upmix of the transmitted downmix derived by application of the transmitted stereo parameters 20. Optionally, a downmix residual 58 is also included in the bitstream. In this case, the residual is decoded and included in the upmix calculation by "stereo decoding" 146.

[0050] The FD path 106 is configured to have its own independent internal joint stereo or multi-channel decoding. For joint stereo, the decoding reuses its own critically sampled, real-valued filter bank 152, e.g., IMDCT.

[0051] The LPD stereo output and the FD stereo output are mixed in the time domain, for example using the first combiner 112, to provide the final output 118 of the fully switched encoder.

[0052] Even though multi-channel is described for stereo decoding in the relevant figures, the same principles apply also to multi-channel processing, generally with more than two channels.

[0053] Fig. 8 shows a schematic block diagram of a method 800 for encoding a multi-channel signal. The method 800 comprises a step 805 of performing linear predictive domain encoding, a step 810 of performing frequency domain encoding, and a step 815 of switching between linear predictive domain encoding and frequency domain encoding. The linear predictive domain encoding comprises a step of downmixing the multi-channel signal to obtain a downmix signal, a linear predictive domain core encoding the downmix signal, and a first joint multi-channel encoding for generating a first multi-channel information from the multi-channel signal. The frequency domain encoding comprises a step of a second joint multi-channel encoding for generating a second multi-channel information from the multi-channel signal. The second joint multi-channel encoding is different from the first multi-channel encoding. The switching is performed such that a portion of the multi-channel signal is represented either by a linear predictive domain encoded frame or by a frequency domain encoded frame.

[0054] Fig. 9 shows a schematic block diagram of a method 900 for decoding an encoded audio signal. The method 900 includes a linear predictive domain decoding step 905, a frequency domain decoding step 910, a first joint multi-channel decoding step 915 for generating a first multi-channel representation using an output of the linear predictive domain decoding and the first multi-channel information, a second multi-channel decoding step 920 for generating a second multi-channel representation using an output of the frequency domain decoding and the second multi-channel information, and a step 925 for combining the first multi-channel representation and the second multi-channel representation to obtain a decoded audio signal. The second first multi-channel information decoding step is different from the first multi-channel decoding step.

[0055] Fig. 10 shows a schematic block diagram of an audio encoder for encoding a multi-channel signal according to another embodiment. The audio encoder 2' includes a linear prediction domain encoder 6 and a multi-channel residual encoder 56. The linear prediction domain encoder includes a downmixer 12 for downmixing the multi-channel signal 4 to obtain a downmix signal 14, and a linear prediction domain core encoder 16 for encoding the downmix signal 14. The linear prediction domain encoder 6 further includes a joint multi-channel encoder 18 for generating multi-channel information 20 from the multi-channel signal 4. The linear prediction domain encoder further includes a linear prediction domain decoder 50 for decoding the encoded downmix signal 26 to obtain an encoded and decoded downmix signal 54. The multi-channel residual encoder 56 uses the encoded and decoded downmix signal 54 to calculate and encode a multi-channel residual signal. The multi-channel residual signal represents an error between the decoded multi-channel representation 54 using the multi-channel information 20 and the multi-channel signal 4 before downmixing.

[0056] According to an embodiment, the downmix signal 14 comprises a low band and a high band. The linear prediction domain decoder uses a bandwidth extension processor to apply a bandwidth extension process to the parametric coding of the high band. The linear prediction domain encoder is configured to obtain only a low band signal representing the low band of the downmix signal as the coded and decoded downmix signal 54. The coded multi-channel residual signal only has a band corresponding to the low band of the multi-channel signal before the downmix. Furthermore, the same explanations as for the audio encoder 2 apply to the audio encoder 2'. However, the separate frequency coding of the encoder 2 is omitted. This simplifies the encoder configuration and is therefore advantageous if the encoder is used for an audio signal that only contains a signal that is parametrically coded in the time domain without noticeable quality loss or the quality of the decoded audio signal is still within the standard. However, a dedicated residual stereo coding is advantageous to increase the reconstruction quality of the decoded audio signal. More specifically, since the differences of the decoded audio signal relative to the encoded audio signal are known by the decoder, the differences between the audio signal before encoding and the encoded and decoded audio signal are derived and transmitted to the decoder in order to increase the reproduction quality of the decoded audio signal.

[0057] FIG. 11 shows an audio decoder 102' for decoding an encoded audio signal 103 according to another embodiment. The audio decoder 102' includes a linear prediction domain decoder 104 and a combined multi-channel decoder 108 for generating a multi-channel representation 114 using the output of the linear prediction domain decoder 104 and the combined multi-channel information 20. Furthermore, the encoded audio signal 103 includes a multi-channel residual signal 58 that is used by the multi-channel decoder for generating the multi-channel representation 114. Furthermore, the same explanations related to the audio decoder 102 apply to the audio decoder 102'. Here, even if a parametric and therefore wasteful coding is used, the residual signal from the original audio signal to the decoded audio signal is used and applied to the decoded audio signal to achieve at least approximately the same quality of the decoded audio signal compared to the original audio signal. However, the frequency decoding part shown with respect to the audio decoder 102 is omitted in the audio decoder 102'.

[0058] Fig. 12 shows a schematic block diagram of an audio encoding method 1200 for encoding a multi-channel signal. The method 1200 comprises a step 1205 of linear predictive domain encoding comprising a downmix of the multi-channel signal to obtain a downmixed multi-channel signal. A linear predictive domain core encoder generates multi-channel information from the multi-channel signal. The method further comprises linear predictive domain decoding the downmix signal to obtain an encoded and decoded downmix signal. The method 1200 comprises a step 1210 of multi-channel residual encoding, using the encoded and decoded downmix signal to calculate an encoded multi-channel residual signal. The multi-channel residual signal represents the error between the decoded multi-channel representation using the first multi-channel information and the multi-channel signal before the downmix.

[0059] Fig. 13 shows a schematic block diagram of a method 1300 for decoding an encoded audio signal. The method 1300 comprises a step 1305 of linear predictive domain decoding and a step 1310 of joint multi-channel decoding using an output of the linear predictive domain decoding and the joint multi-channel information to generate a multi-channel representation. The encoded multi-channel audio signal comprises channel residual signals. The joint multi-channel decoding uses the multi-channel residual signals to generate the multi-channel representation.

[0060] The described embodiments allow for use in the distribution of all types of broadcast stereo or multi-channel audio content (speech and music alike with constant perceptual quality at a given low bit rate), such as, for example, digital radio, Internet streaming and audio communication applications.

[0061] 14 to 17 illustrate an embodiment of how to apply the proposed seamless switching between LPD coding and frequency domain coding, and vice versa. In general, the past windowing or processing is shown using thin lines, and the thick lines indicate the current windowing or processing. The switching is applied, and the dotted lines indicate the current processing that is done exclusively for the transition or switch. Switching or transition from LPD coding to frequency coding.

[0062] FIG. 14 shows an overview timing diagram illustrating an embodiment for seamless switching between frequency domain coding and time domain coding. This is appropriate if, for example, the controller 10 indicates that the current frame would be better coded using LPD coding instead of the FD coding used for the previous frame. During frequency domain coding, stopping windows 200a and 200b are applied to each stereo signal (optionally extended to two or more channels). The stopping window differs from the standard MDCT overlap-add fading at the beginning 202 of the first frame 204. The left part of the stopping window is the traditional overlap-add for coding the previous frame, for example using the MDCT time-frequency transform. Thus, the frames before the switch are still properly coded. When the switch is applied to the current frame 204, additional stereo parameters are calculated even though the first parametric representation of the intermediate signal for the time domain coding is calculated for the following frame 206. These two additional stereo analyses are made to be able to generate the intermediate signal 208 for the LPD look-ahead. However, stereo parameters are (additionally) transmitted for the two first LPD stereo windows. In the normal case, the stereo parameters are sent with two LPD stereo frames of delay. The intermediate signal is also used for past updates of the ACELP memory, such as LPC analysis or forward aliasing cancellation (FAC). Later, the LPD stereo windows 210a-210d for the first stereo signal and the LPD stereo windows 212a-212d for the second stereo signal are applied in the analysis filter bank 82 before applying a time-frequency transformation, for example using a DFT. The intermediate signal includes a typical cross-fade gradient when using TCX coding, resulting in an exemplary LPD analysis window 214. If ACELP is used to code an audio signal, such as a mono low-band signal, it simply selects a number of frequency bands, indicated by a rectangular LPD analysis window 216, to which the LPC analysis is applied.

[0063] Furthermore, the timing indicated by the vertical line 218 indicates that the current frame to which the transition is applied contains information from the frequency domain analysis window 200a, 200b and the calculated intermediate signal 208 as well as the corresponding stereo information. During the horizontal part of the frequency analysis window between the line 202 and the line 218, the frame 204 is completely coded using frequency domain coding. From the line 218 to the end of the frequency analysis window at the line 220, the frame 204 contains information from both frequency domain coding and LPD coding, and from the line 220 to the end of the frame 204 at the vertical line 222, only LPD coding contributes to the coding of the frame. More attention is drawn to the middle part of the coding, since the first and the last (third) part are easily derived from one coding technique without aliasing. However, for the middle part, it should be distinguished between ACELP and TCX mono signal coding. Since TCX coding uses cross-fading, as already applied by frequency domain coding, a simple fade out of the frequency coded signal and a fade in of the TCX coded intermediate signal provides complete information for coding the current frame 204. If ACELP is used for mono signal coding, the area 224 does not contain complete information for coding the audio signal, so more sophisticated processing is applied. The proposed method is for example the Forward Aliasing Correction (FAC) described in the USAC standard in section 7.16.

[0064] According to an embodiment, the controller 10 is configured to switch within a current frame 204 of the multi-channel audio signal from using a frequency domain encoder 8 for encoding previous frames to a linear prediction domain encoder for decoding subsequent frames. The first joint multi-channel encoder 18 calculates synthesis multi-channel parameters 210a, 210b, 212a, 212b from the multi-channel audio signal for the current frame. The second joint multi-channel encoder 22 is configured to weight the second multi-channel signal using a stopping window.

[0065] Fig. 15 shows a schematic timing diagram of the decoder corresponding to the encoder operation of Fig. 14. Here, the reconstruction of the current frame 204 is described according to an embodiment. As already shown in the encoder timing diagram of Fig. 14, the frequency domain stereo channels are provided from the frame before applying the stop windows 200a and 200b. The transition from FD to LPD mode is first made on the decoded intermediate signal, as in the mono case. It is achieved by artificially creating an intermediate signal 226 from the decoded time domain signal 116 in FD mode. ccfl is the core code frame length and L_fac denotes the length of the frequency aliasing cancellation window or frame or block or transform.

[0066] TIFF0007673035000001.tif11167

[0067] This signal is then transmitted to the LPD decoder 120 for updating the memory and applying the FAC decoding as it would be done in the mono case for the transition from FD mode to ACELP. The process is described in the USAC standard [ISO / IEC DIS 23003-3, USAC] in section 7.16. In the case from FD mode to TCX, a conventional overlap-and-add is performed. The LPD stereo decoder 146 receives as input signal the intermediate signal decoded (in the frequency domain after the time-frequency conversion of the time-frequency converter 144 has been applied) by applying, for example, the transmitted stereo parameters 210 and 212 to the stereo processing already transitioned. The stereo decoder then outputs left and right channel signals 228, 230, which overlap with the previous frame decoded in FD mode. The signals, i.e. the FD decoded time domain signal and the LPD decoded time domain signal for the frame to which the transition is applied, are then cross-faded (in the combiner 112) in the respective channels to smooth the transition in the left and right channels.

[0068] TIFF0007673035000002.tif74162

[0069] In Fig. 15 the transition is illustrated diagrammatically with M = ccfl / 2. Furthermore, the combiner performs a cross-fade on consecutive frames being decoded using only FD or LPD decoding, without transitioning between these modes.

[0070] That is, the overlap-add process of FD decoding is replaced by a cross-fade of the FD decoded audio signal and the LPD decoded audio signal, especially when using MDCT / IMDCT for time-frequency / frequency-time transformation. Thus, the decoder should calculate an LPD signal for the fade-out part of the FD decoded audio signal in order to fade in the LPD decoded audio signal. According to an embodiment, the audio decoder 102 is configured to switch from using the frequency domain decoder 106 for decoding the previous frame to the linear prediction domain decoder 104 for decoding the subsequent frame within a current frame 204 of the multi-channel audio signal. The combiner 112 calculates a synthetic intermediate signal 226 from the second multi-channel representation 116 of the current frame. The first combined multi-channel decoder 108 uses the synthetic intermediate signal 226 and the first multi-channel information 20 to generate the first multi-channel representation 114. Furthermore, the combiner 112 is configured to combine the first multi-channel representation and the second multi-channel representation to obtain a decoded current frame of the multi-channel audio signal.

[0071] FIG. 16 shows a schematic timing diagram in the encoder for performing the transition from using LPD coding to using FD decoding in the current frame 232. To switch from LPD coding to FD coding, start windows 300a, 300b are applied to the FD multi-channel coding. The start windows have a similar function when compared to the stop windows 200a, 200b. During the fade-out of the TCX coded mono signal in the LPD encoder between the vertical lines 234 and 236, the start windows 300a, 300b perform a fade-in. When using ACELP instead of TCX, the mono signal does not perform a smooth fade-out. Nevertheless, the correct audio signal is reconstructed in the decoder using, for example, FAC. The LPD stereo windows 238 and 240 are calculated by default and refer to the ACELP or TCX coded mono signal and are indicated by the LPD analysis window 241.

[0072] FIG. 17 shows a schematic timing diagram for a decoder that corresponds to the encoder timing diagram described with reference to FIG.

[0073] For the transition from LPD mode to FD mode, a special frame is decoded by the stereo decoder 146. The intermediate signal coming from the LPD mode decoder is extended with zeros for frame index i=ccfl / M.

[0074] TIFF0007673035000003.tif15162

[0075] The stereo decoding previously described is performed by retaining the last stereo parameters and switching off the side signal inverse quantization, i.e. code_mode is set to 0. Furthermore, no right windowing after the inverse DFT is applied, which results in sharp edges 242a, 242b of the special LPD stereo windows 244a, 244b. It is clearly observed that edges of a specific shape are placed in flat sections 246a, 246b. The entire information of the corresponding part of the frame is derived from the FD encoded audio signal. Thus, right windowing (without sharp edges) would result in unwanted interference from the LPD information to the FD information and is therefore not applied.

[0076] The resulting left and right (decoded LPD) channels 250a, 250b (using the LPD decoded intermediate signal indicated by the LPD analysis window 248 and the stereo parameters) are then combined into the FD mode decoded channels of the next frame by using overlap-add processing in the TCX to FD mode case, or by using FAC per channel in the ACELP to FD mode case. A schematic illustration of the transition is given in FIG. 17, where M=ccfl / 2.

[0077] According to an embodiment, the audio decoder 102 switches from using the linear prediction domain decoder 104 for decoding previous frames to the frequency domain decoder 106 for decoding subsequent frames within a current frame 232 of the multi-channel audio signal. The stereo decoder 146 calculates a composite multi-channel audio signal from the decoded mono signal of the linear prediction domain decoder for the current frame using the multi-channel information of the previous frame. The second combined multi-channel decoder 110 calculates a second multi-channel representation for the current frame and weights the second multi-channel representation using a starting window. The combiner 112 combines the composite multi-channel audio signal and the weighted second multi-channel representation to obtain the decoded current frame of the multi-channel audio signal.

[0078] Fig. 18 shows a schematic block diagram of an encoder 2'' for encoding a multi-channel signal 4. The audio encoder 2'' comprises a downmixer 12, a linear prediction domain core encoder 16, a filter bank 82 and a joint multi-channel encoder 18. The downmixer 12 is configured to downmix the multi-channel signal 4 to obtain a downmix signal 14. The downmix signal is a mono signal, e.g. an intermediate signal of an M / S multi-channel audio signal. The linear prediction domain core encoder 16 encodes the downmix signal 14. The downmix signal 14 has a low band and a high band. The linear prediction domain core encoder 16 is configured to apply a bandwidth extension process and to apply parametric coding of the high band. Furthermore, the filter bank 82 generates a spectral representation of the multi-channel signal 4. The joint multi-channel encoder 18 is configured to process the spectral representation comprising the low band and the high band of the multi-channel signal to generate multi-channel information 20. The multi-channel information includes ILD and / or IPD and / or IID (Inter-Perception Intensity Difference) parameters that enable the decoder to recalculate the multi-channel audio signal from the mono signal. A more detailed illustration of another aspect of an embodiment according to this aspect can be seen in the previous figures, in particular in FIG.

[0079] According to an embodiment, the linear prediction domain core encoder 16 further comprises a linear prediction domain decoder for decoding said encoded downmix signal 26 to obtain an encoded and decoded downmix signal 54, whereby the linear prediction domain core encoder forms an intermediate signal of the M / S audio signal to be encoded for transmission to a decoder. Furthermore, the audio encoder further comprises a multi-channel residual encoder 56 for calculating an encoded multi-channel residual signal 58 using the encoded and decoded downmix signal 54. The multi-channel residual signal represents the error between the decoded multi-channel representation and the multi-channel signal 4 before downmix using the multi-channel information 20. In other words, the multi-channel residual signal 58 is a side signal of the M / S audio signal and corresponds to the intermediate signal calculated using the linear prediction domain core encoder.

[0080] According to another embodiment, the linear predictive domain core encoder 16 is configured to apply a bandwidth extension process to parametrically code the high band and obtain only a low band signal representing the low band of the downmix signal as the coded and decoded downmix signal. The coded multi-channel residual signal 58 only has a band corresponding to the low band of the multi-channel signal before the downmix. Additionally or alternatively, the multi-channel residual encoder simulates the time domain bandwidth extension applied to the high band of the multi-channel signal in the linear predictive domain core encoder and calculates a residual or side signal for the high band to allow a more accurate decoding of the mono or intermediate signal to derive the decoded multi-channel audio signal. The simulation includes the same or similar calculations performed in the decoder to decode the bandwidth extension high band. An alternative or additional approach to simulate the bandwidth extension is the prediction of the side signal. Thus, the multi-channel residual encoder calculates a full-band residual signal from a parametric representation 83 of the multi-channel audio signal 4 after a time-frequency transformation in a filter bank 82. This all-band side signal is compared with a frequency representation of the all-band intermediate signal, which is also derived from the parametric representation 83. The all-band intermediate signal is calculated, for example, as the sum of the left and right channels of the parametric representation 83, and the all-band side signal is calculated as the difference therefrom. Thus, a further prediction is calculated, a prediction factor of the all-band intermediate signal that minimizes the absolute difference of the all-band side signal, and the product of the prediction factor and the all-band intermediate signal.

[0081] That is, the linear prediction domain encoder is configured to calculate the downmix signal 14 as a parametric representation of the intermediate signal of the M / S multi-channel audio signal. The multi-channel residual encoder is configured to calculate a side signal corresponding to the intermediate signal of the M / S multi-channel audio signal. The residual encoder calculates the high band of the intermediate signal using a simulated time domain bandwidth expansion. Alternatively, the residual encoder predicts the high band of the intermediate signal using a finding of prediction information that minimizes the difference between the calculated side signal and the calculated full-band intermediate signal from a previous frame.

[0082] Another embodiment shows a linear prediction domain core encoder 16 including an ACELP processor 30. The ACELP processor operates on a downsampled downmix signal 34. Furthermore, a time domain bandwidth extension processor 36 is configured to parametrically code the bands of the part of the downmix signal that has been removed from the ACELP input signal by the third downsampling. Additionally or alternatively, the linear prediction domain core encoder 16 includes a TCX processor 32. The TCX processor 32 operates on a downmix signal 14 that is not downsampled or that has been downsampled to a lesser extent than the downsampling for the ACELP processor. Furthermore, the TCX processor includes a first time-to-frequency converter 40, a first parameter generator 42 for generating a parametric representation 46 of the first set of bands, and a first quantizer encoder 44 for generating a set of quantized and coded spectral lines 48 for the second set of bands. The ACELP and TCX processors may either operate separately, e.g., a first number of frames are encoded using ACELP and a second number of frames are encoded using TCX, or both ACELP and TCX contribute information to the decoding of one frame in a joint manner.

[0083] Another embodiment shows a time-frequency converter 40 different from the filter bank 82. The filter bank 82 contains filter parameters optimized for generating a spectral representation 83 of the multi-channel signal 4. The time-frequency converter 40 contains filter parameters optimized for generating a parametric representation 46 of the first set of bands. It should be noted that in another step, the linear prediction domain encoder uses a different filter bank or even no filter bank in case of bandwidth extension and / or ACELP. Furthermore, the filter bank 82 calculates separate filter parameters for generating the spectral representation 83 without relying on the previous parameter selection of the linear prediction domain encoder. That is, the multi-channel coding in LPD mode uses a filter bank for multi-channel processing (DFT) that is not the one used in the bandwidth extension (time domain for ACELP and MDCT for TCX). The advantage is that each parametric coding can use its optimal time-frequency decomposition to obtain its parameters. For example, a combination of ACELP+TDBWE and a parametric multi-channel coding with an external filter bank (e.g. DFT) is advantageous. This combination is particularly efficient since it is known that the best bandwidth extension for speech is in the time domain and multichannel processing is in the frequency domain. Since ACELP+TDBWE does not have any time-to-frequency converters, external filter banks or transforms such as DFT are preferred or even necessary. Other concepts always use the same filter bank and therefore do not use different filter banks, e.g. -MDCT AAC vs. IGF and joint stereo coding -SBR+PS for HeAACv2 in QMF -SBR+MPS212 against QMF's USAC.

[0084] According to another embodiment, the multi-channel encoder comprises a first frame generator and the linear prediction domain core encoder comprises a second frame generator. The first and second frame generators are configured to form frames from the multi-channel signal 4. The first and second frame generators are configured to form frames of equal length, i.e. the framing of the multi-channel processor is the same as that used in ACELP. Even if the multi-channel processing is done in the frequency domain, the time resolution for calculating its parameters or downmix is ​​close to or even equal to the framing of ACELP. The equal length in this case relates to the framing of ACELP being equal or close to the time resolution for calculating the parameters for the multi-channel processing or downmix.

[0085] According to another embodiment, the audio encoder further comprises a linear prediction domain encoder 6 including a linear prediction domain core encoder 16 and a multi-channel encoder 18, a frequency domain encoder 8 and a controller 10 for switching between the linear prediction domain encoder 6 and the frequency domain encoder 8. The frequency domain encoder 8 comprises a second combined multi-channel encoder 22 for encoding second multi-channel information 24 from the multi-channel signal. The second combined multi-channel encoder 22 is different from the first combined multi-channel encoder 18. Furthermore, the controller 10 is configured such that portions of the multi-channel signal are represented either by encoded frames of the linear prediction domain encoder or by encoded frames of the frequency domain encoder.

[0086] FIG. 19 shows a schematic block diagram of a decoder 102″ for decoding an encoded audio signal 103 including a core encoded signal, a bandwidth extension parameter and multi-channel information according to another embodiment. The audio decoder includes a linear prediction domain core decoder 104, an analysis filter bank 144, a multi-channel decoder 146 and a synthesis filter bank processor 148. The linear prediction domain core decoder 104 decodes the core encoded signal to generate a mono signal, which is a (full-band) intermediate signal of the M / S encoded audio signal. The analysis filter bank 144 converts the mono signal to a spectral representation 145. The multi-channel decoder 146 generates a first channel spectrum and a second channel spectrum from the spectral representation of the mono signal and the multi-channel information 20. Thus, the multi-channel decoder uses the multi-channel information including, for example, a side signal corresponding to the decoded intermediate signal. The synthesis filter bank processor 148 is configured to synthesize filter the first channel spectrum to obtain the first channel signal and to synthesize filter the second channel spectrum to obtain the second channel signal. Thus, preferably, the reverse operations compared to the analysis filter bank 144 are applied to the first and second channel signals, which are IDFT if the analysis filter bank uses DFT. However, the filter bank processor processes the two channel spectra, for example in parallel or in successive order, for example using the same filter bank. A more detailed view of this alternative embodiment can be seen in relation to the previous figures, in particular FIG. 7.

[0087] According to another embodiment, the linear prediction domain core decoder comprises a bandwidth extension processor 126 for generating a highband part 140 from the bandwidth extension parameter and the lowband mono signal or the core coded signal to obtain a decoded highband 140 of the audio signal. The lowband signal processor is configured to decode the lowband mono signal. The combiner 128 is configured to calculate a fullband mono signal using the decoded lowband mono signal of the audio signal and the decoded highband of the audio signal. The lowband mono signal is for example a baseband representation of an intermediate signal of an M / S multi-channel audio signal. The bandwidth extension parameter is applied to calculate (in the combiner 128) the fullband mono signal from the lowband mono signal.

[0088] According to another embodiment, the linear prediction domain decoder includes an ACELP decoder 120, a low-band synthesizer 122, an upsampler 124, a time-domain bandwidth extension processor 126, or a second combiner 128. The second combiner 128 is configured to combine the upsampled low-band signal and the bandwidth-extended high-band signal 140 to obtain a full-band ACELP decoded mono signal. The linear prediction domain decoder further includes a TCX decoder 130 and an intelligent gap filling processor 132 to obtain a full-band TCX decoded mono signal. Thus, the full-band synthesis processor 134 combines the full-band ACELP decoded mono signal and the full-band TCX decoded mono signal. Furthermore, a cross-path 136 is provided to initialize the low-band synthesizer with information derived by the low-band spectro-temporal transformation from the TCX decoder and the IGF processor.

[0089] According to another embodiment, the audio decoder comprises a frequency domain decoder 106, a second combined multi-channel decoder 110 for generating a second multi-channel representation 116 using the output 22 of the frequency domain decoder 106 and the second multi-channel information 24, and a first combiner 112 for combining the first and second channel signals to the second multi-channel representation 116 to obtain a decoded audio signal 118. The second combined multi-channel decoder is different from the first combined multi-channel decoder. Thus, the audio decoder switches between parametric multi-channel decoding using LPD or frequency domain decoding. This approach has already been described in detail with respect to the previous figures.

[0090] According to another embodiment, the analysis filterbank 144 includes a DFT to transform the mono signal into a spectral representation 145. The fullband synthesis processor 148 includes an IDFT to transform the spectral representation 145 into the first and second channel signals. Furthermore, the analysis filterbank applies a window to the DFT-transformed spectral representation 145 such that the previous and current frames are consecutive and the right part of the spectral representation of the previous frame overlaps with the left part of the spectral representation of the current frame. That is, a cross-fade is applied from one DFT block to another DFT block to perform a smooth transition between consecutive DFT blocks and / or to reduce blocking artifacts.

[0091] According to another embodiment, the multi-channel decoder 146 is configured to derive the first and second channel signals from the mono signal. The mono signal is an intermediate signal of the multi-channel signal. The multi-channel decoder 146 is configured to derive an M / S multi-channel decoded audio signal. The multi-channel decoder 146 is configured to calculate a side signal from the multi-channel information. Furthermore, the multi-channel decoder 146 is configured to calculate an L / R multi-channel decoded audio signal from the M / S multi-channel decoded audio signal. The multi-channel decoder 146 calculates an L / R multi-channel decoded audio signal for the low band using the multi-channel information and the side signal. Additionally or alternatively, the multi-channel decoder 146 calculates a predicted side signal from the intermediate signal. The multi-channel decoder is further configured to calculate an L / R multi-channel decoded audio signal for the high band using the predicted side signal and an ILD value of the multi-channel information.

[0092] Moreover, the multi-channel decoder 146 is further configured to perform a complex operation on the L / R decoded multi-channel audio signal. The multi-channel decoder uses the energy of the encoded intermediate signal and the energy of the decoded L / R multi-channel audio signal to calculate the magnitude of the complex operation to obtain energy compensation. Furthermore, the multi-channel decoder is configured to calculate the phase of the complex operation using the IPD value of the multi-channel information. After decoding, the energy, level or phase of the decoded multi-channel signal is different from the decoded mono signal. Therefore, the complex operation is determined such that the energy, level or phase of the multi-channel signal is adapted to the value of the decoded mono signal. Furthermore, the phase is adapted to the value of the phase of the multi-channel signal before encoding, for example, using the IPD parameter calculated from the multi-channel information calculated on the encoder side. Furthermore, the human perception of the decoded multi-channel signal is adapted to the human perception of the original multi-channel signal before encoding.

[0093] Fig. 20 shows a schematic representation of a flow chart of a method 2000 for encoding a multi-channel signal. The method comprises a step 2050 of downmixing the multi-channel signal to obtain a downmix signal and a step 2100 of encoding the downmix signal. The downmix signal has a low band and a high band. A linear prediction domain core encoder is configured to parametrically encode the high band by applying a bandwidth extension process. Furthermore, the method comprises a step 2150 of generating a spectral representation of the multi-channel signal and a step 2200 of processing the spectral representation comprising the low band and the high band of the multi-channel signal to generate multi-channel information.

[0094] Figure 21 shows a schematic flow chart of a method 2100 for decoding an encoded audio signal, including a core encoded signal, a bandwidth extension parameter and multi-channel information. The method comprises a step 2105 of decoding the core encoded signal to generate a mono signal, a step 2110 of converting the mono signal to a spectral representation, a step 2115 of generating a first channel spectrum and a second channel spectrum from the spectral representation of the mono signal and the multi-channel information, a step of synthesizing filtering the first channel spectrum to obtain the first channel signal, and a step 2120 of synthesizing filtering the second channel spectrum to obtain the second channel signal.

[0095] Another embodiment is described as follows.

[0096] Bitstream Syntax Changes Section 5.3.2, Table 23 of the USAC Specification [1] for Auxiliary Payloads, should be amended as follows:

[0097] [Table 1]

[0098] The following table should be added:

[0099] [Table 2]

[0100] The following payload description should be added to Section 6.2, USAC Payloads:

[0101] 6.2.x lpd_stereo_stream() The detailed decoding procedure is explained in the 7.x LPD Stereo Decoding section.

[0102] Terms and Definitions lpd_stereo_stream(): Data element for decoding stereo data for LPD mode. res_mode: A flag indicating the frequency resolution of the parameter band. q_mode: A flag indicating the time resolution of the parameter band. ipd_mode: A bit field that defines the maximum parameter band for the IPD parameters. pred_mode: A flag indicating if prediction is used. cod_mode: A bitfield that defines the maximum value of the parameter band for which the side signal is quantized. Ild_idx[k][b]: The ILD parameter index for frame k and band b. Ipd_idx[k][b]: IPD parameter index for frame k and band b. pred_gain_idx[k][b]: Prediction gain index for frame k and band b. cod_gain_idx: Global gain index for the quantized side signal.

[0103] Auxiliary elements ccfl: Core Code Frame Length. M: Stereo LPD frame length as defined in Table 7.x.1. band_config(): A function that returns the number of encoded parameter bands. The function is defined in 7.x. band_limits(): A function that returns the number of encoded parameter bands. The function is defined in 7.x. max_band(): A function that returns the number of encoded parameter bands. The function is defined in 7.x. ipd_max_band(): A function that returns the number of encoded parameter bands. The function is defined in 7.x. cod_max_band(): A function that returns the number of coded parameter bands. The function is defined in 7.x. cod_L: The number of DFT lines for the decoded side signal.

[0104] Decryption Process LPD Stereo Coding Tool Description LPD stereo is a discrete M / S stereo coding. The middle channel is coded by a mono LPD core coder and the side signal is coded in the DFT domain. The decoded middle signal is output from the LPD mono decoder and then processed by the LPD stereo module. The stereo decoding is done in the DFT domain where the L and R channels are decoded. The two decoded channels are transformed back in the time domain and then combined in this domain with the decoded channel from the FD mode. The FD coding mode uses its own stereo tools, i.e. discrete stereo, with or without complex prediction.

[0105] Data Elements res_mode: A flag indicating the frequency resolution of the parameter band. q_mode: A flag indicating the time resolution of the parameter band. ipd_mode: A bit field that defines the maximum parameter band for the IPD parameters. pred_mode: A flag indicating if prediction is used. cod_mode: A bitfield that defines the maximum value of the parameter band for which the side signal is quantized. Ild_idx[k][b]: The ILD parameter index for frame k and band b. Ipd_idx[k][b]: IPD parameter index for frame k and band b. pred_gain_idx[k][b]: Prediction gain index for frame k and band b. cod_gain_idx: Global gain index for the quantized side signal.

[0106] Auxiliary elements ccfl: Core Code Frame Length. M: Stereo LPD frame length as defined in Table 7.x.1. band_config(): A function that returns the number of encoded parameter bands. The function is defined in 7.x. band_limits(): A function that returns the number of encoded parameter bands. The function is defined in 7.x. max_band(): A function that returns the number of encoded parameter bands. The function is defined in 7.x. ipd_max_band(): A function that returns the number of encoded parameter bands. The function is defined in 7.x. cod_max_band(): A function that returns the number of coded parameter bands. The function is defined in 7.x. cod_L: The number of DFT lines for the decoded side signal.

[0107] Decryption Process Stereo decoding is performed in the frequency domain. It works as a post-processing of the LPD decoder. It receives the synthesis of the mono intermediate signal from the LPD decoder. The side signal is then decoded or predicted in the frequency domain. The channel spectrum is then reconstructed in the frequency domain before being resynthesized in the time domain. Stereo LPD works with a fixed frame length equal to the size of the ACELP frame, independent of the coding mode used in the LPD mode.

[0108] Frequency Analysis The DFT spectrum of frame index i is computed from the decoded frame x of length M.

[0109] TIFF0007673035000006.tif16124 where N is the size of the analysis of the signal. w is the analysis window. x is the decoded time signal from the LPD decoder at frame index i delayed by the DFT overlap size L. M is the sampling rate used in FD mode and is equal to the size of the ACELP frame. N is equal to the stereo LPD frame size plus the DFT overlap size. The size depends on the LPD version used, as reported in Table 7.x.1.

[0110] [Table 3]

[0111] TIFF0007673035000008.tif37166

[0112] Parameter band configuration The DFT spectrum is divided into non-overlapping frequency bands called parameter bands. The spectral partitioning is non-uniform and resembles the frequency decomposition of hearing. Two different partitionings of the spectrum are possible with subsequent bandwidths of approximately two or four times the equivalent rectangular bandwidth (ERB). The spectral partitioning is selected by the data element res_mod and is defined by the following pseudocode: funtion nbands=band_config(N,res_mod) band_limits[0]=1; nbands=0; while(band_limits[nbands++]<(N / 2))[ if(stereo_lpd_res==0) band_limits[nbands]=band_limits_erb2[nbands]; else band_limits[nbands]=band_limits_erb4[nbands]; ] nbands--; band_limits[nbands]=N / 2; return nbands where nbands is the total number of parameter bands and N is the DFT analysis window size. The tables band_limits_erb2 and band_limits_erb4 are defined in Table 7.x.2. The decoder can adaptively change the resolution of the spectral parameter bands in every two stereo LPD frames.

[0113] [Table 4]

[0114] The maximum number of parameter bands for the IPD is signaled in the 2-bit field ipd_mod data element. ipd_max_band=max_band[res_mod][ipd_mod] The maximum number of parameter bands for the coding of the side signal is signaled in the 2-bit field cod_mod data element. cod_max_band=max_band[res_mod][cod_mod] The table max_band[][] is defined in Table 7.x.3. To predict the side signal, the number of decoded lines is then calculated as follows: cod_L=2·(band_limits[cod_max_band]-1)

[0115] [Table 5]

[0116] Inverse quantization of stereo parameters The stereo parameter inter-channel level difference (ILD), inter-channel phase difference (IPD), and prediction gain are sent to all frames or all pairs of frames that depend on the flag q_mode. If q_mode is equal to 0, all parameters are updated for all frames. Otherwise, the parameter values are updated only for the odd indices i of the stereo LPD frames within the USAC frame. The index i of the stereo LPD frames within the USAC frame can be either between 0 and 3 in LPD version 0, or between 0 and 1 in LPD version 1.

[0117] ILD is decoded as follows. For 0 ≦ b < nbands, ILD i [b] = ild_q[ild_idx[i][b]]

[0118] TIFF0007673035000011.tif16140

[0119] TIFF0007673035000012.tif28169

[0120] TIFF0007673035000013.tif33168

[0121] TIFF0007673035000014.tif21170

[0122]

Table 6

[0123]

Table 7

[0124] TIFF0007673035000017.tif54170

[0125] TIFF0007673035000018.tif62170

[0126] TIFF0007673035000019.tif81169

[0127] TIFF0007673035000020.tif56170

[0128] TIFF0007673035000021.tif33166

[0129] Post-processing Bass post-processing is done separately on the two channels. The processing is the same for both channels as described in section 7.17 of [1].

[0130] It should be understood that in this specification, signals on a line are sometimes named by the reference number of the line, and sometimes are indicated by the reference number attributed to the line itself. Thus, the notation is such that a line carrying a signal indicates the signal itself. A line can be a physical line in a hardwired implementation. However, in a computerized implementation, there is no physical line, but the signal represented by the line is transmitted from one computing module to another.

[0131] Although the invention has been described in the context of block diagrams, where the blocks represent actual or logical hardware components, the invention may also be practiced by a computer-implemented method, where the blocks represent corresponding method steps, and these steps represent functions performed by corresponding logical or physical hardware blocks.

[0132] Even if some aspects are described in the context of an apparatus, the aspects are also understood as expressing a description of the corresponding method. As a result, a block or apparatus may correspond to a method step or be understood as a feature of a method step. By analogy, the aspects described therewith or the method steps also correspond to blocks or represent a description of corresponding details or characteristics of the apparatus. Some or all of the method steps may be performed by (or in conjunction with) a hardware apparatus, e.g., a microprocessor, a programmable computer, or an electronic circuit. In some embodiments, some or some of the most important method steps may be performed by such an apparatus.

[0133] The transmitted or encoded signals of the present invention may be stored on a digital storage medium or transmitted over a transmission medium such as a wireless transmission medium or a wired transmission medium such as the Internet.

[0134] Depending on specific implementation requirements, embodiments of the present invention may be implemented in hardware or in software. The implementation may be implemented using a digital storage medium having electronically readable control signals stored thereon, such as a floppy disk, DVD, Blu-ray disk, CD, ROM, PROM, EPROM, EEPROM, or FLASH memory, which may cooperate with a programmable computer system so that the respective methods are executed. Thus, the digital storage medium may be computer readable.

[0135] Some embodiments of the present invention include a data carrier having electronically readable control signals that can cooperate with a programmable computer system to cause some of the methods described herein to be performed.

[0136] Typically, embodiments of the invention are implemented as a computer program product having program code, which, when the computer program product runs on a computer, is operated to perform a certain method, the program code being, for example, stored on a machine readable carrier.

[0137] Other embodiments comprise the computer program for performing some of the methods described herein, the computer program being stored on a machine readable carrier.

[0138] In other words, embodiments of the inventive method are, therefore, a computer program having a program code for performing some of the methods described herein, when the computer program runs on a computer.

[0139] A further embodiment of the inventive method is therefore a data carrier (or digital storage medium, or computer readable medium) comprising a computer program for performing some of the methods described herein. The data carrier, digital storage medium or recorded medium is typically a tangible and / or intangible object.

[0140] A further embodiment of the inventive method is, therefore, a data stream or a sequence of signals representing the computer program for performing some of the methods described herein, for example a data stream or a sequence of signals adapted to be transmitted via a data communications connection, for example the Internet.

[0141] A further embodiment comprises a processing means, for example a computer, or a programmable logic circuit, configured to or adapted to perform some of the methods described herein.

[0142] Further embodiments include a computer having installed thereon the computer program for performing some of the methods described herein.

[0143] Another embodiment according to the invention includes an apparatus or system configured to transfer a computer program for performing at least one of the methods described herein to a receiver. The transfer can be, for example, electronic or optical. The receiver can be, for example, a computer or a mobile device or a storage device. The apparatus or system can, for example, include a file server for transferring the computer program to the receiver.

[0144] In some embodiments, a programmable logic circuit (e.g., a Field Programmable Gate Array (FPGA)) may be used to perform some or all of the functions described herein. In some embodiments, the Field Programmable Gate Array may cooperate with a microprocessor to perform some of the methods described herein. In general, the methods are preferably performed by some hardware device.

[0145] The above-described embodiments merely represent examples of the principles of the present invention. It is understood that modifications and variations of the arrangements and details described herein will be apparent to others skilled in the art. It is therefore the intention to be limited only by the scope of the following claims, rather than by the detailed description of the specification presented herein, by means of the description and discussion of the embodiments.

[0146] literature [1] ISO / IEC DIS 23003-3, Usac [2]ISO / IEC DIS 23008-3, 3D Audio

Claims

1. An audio encoder (2'') for encoding a multi-channel signal (4), comprising: a downmixer (12) for downmixing the multi-channel signal (4) to obtain a downmix signal (14), said downmixer being configured as an active downmixer or as a passive downmixer; a linear prediction domain core encoder (16) for encoding the downmix signal (14) to obtain an encoded downmix signal (26), the downmix signal (14) having a low band of the downmix signal (14) and a high band of the downmix signal (14), the linear prediction domain core encoder (16) being configured to parametrically encode the high band of the downmix signal (14) by applying a bandwidth extension process; a filter bank (82) for generating a spectral representation of said multi-channel signal (4); a joint multi-channel encoder (18) configured to process the spectral representation comprising a low band of the multi-channel signal (4) and a high band of the multi-channel signal (4) to generate multi-channel information (20); Including, the linear prediction domain core encoder (16) further comprises a linear prediction domain decoder (50) for decoding the encoded downmix signal (26) to obtain an encoded and decoded downmix signal (54); The audio encoder (2'') further comprises a multi-channel residual coder (56) for calculating, using the encoded and decoded downmix signal (54), an encoded multi-channel residual signal (58), the encoded multi-channel residual signal (58) representing an error between a decoded multi-channel representation obtained using the multi-channel information (20) and the multi-channel signal (4) before downmix, The linear prediction domain decoder (50) is configured to obtain, as the encoded and decoded downmix signal (54), only a low band signal representing the low band of the downmix signal (14), and the encoded multi-channel residual signal (58) has only a band corresponding to the low band of the multi-channel signal (4) before the downmix, or the encoded multi-channel residual signal (58) has a low band corresponding to the low band of the multi-channel signal (4) before the downmix and a high band corresponding to the high band of the multi-channel signal (4) before the downmix.

2. 2. The audio encoder (2'') of claim 1, wherein the linear prediction domain core encoder (16) comprises an ACELP processor (30) configured to operate on a downsampled downmix signal (34) obtained from the downmix signal (14) by a downsampler (35), and a time domain bandwidth extension processor (36) configured to parametrically encode the high band of the downmix signal (14) removed from the downmix signal (14) by the downsampling using the downsampler (35).

3. 3. The audio encoder (2'') of claim 1 or claim 2, wherein the linear prediction domain core encoder (16) comprises a TCX processor (32) configured to operate on the downmix signal (14) that has not been downsampled, the TCX processor comprising a time-to-frequency converter (40), a parameter generator (42) for generating a parametric representation (46) of a first set of bands, and a quantization encoder (44) for generating a set of quantized encoded spectral lines (48) for a second set of bands.

4. An audio encoder (2'') as described in claim 2, wherein the linear prediction domain core encoder (16) includes a TCX processor (32) configured to operate on the downmix signal (14) that has been downsampled to a lesser extent than the downsampling performed by a downsampler (35) for the ACELP processor, the TCX processor including a time-to-frequency converter (40), a parameter generator (42) for generating a parametric representation (46) of a first set of bands, and a quantization encoder (44) for generating a set of quantized and encoded spectral lines (48) for a second set of bands.

5. 5. An audio encoder (2'') according to claim 3 or claim 4, wherein the time-to-frequency converter (40) is different from the filter bank (82), the filter bank (82) comprising filter parameters optimized for generating a spectral representation of the multi-channel signal (4), or the time-to-frequency converter (40) comprises filter parameters optimized for generating the parametric representation (46) of a first set of bands.

6. 6. An audio encoder (2'') according to claim 1, wherein the combined multi-channel encoder (18) comprises a first frame generator and the linear prediction domain core encoder (16) comprises a second frame generator, the first frame generator and the second frame generator being configured to form frames from the multi-channel signal (4), the first frame generator and the second frame generator being configured to form frames of equal length.

7. the active downmixer comprises a time-to-frequency converter, a downmix stage and a frequency-to-time converter, or 7. An audio encoder (2'') according to any of claims 1 to 6, wherein the passive downmixer comprises an adder for adding a first channel of the multi-channel signal (4) and a second channel of the multi-channel signal (4) after using weighting.

8. said linear prediction domain core encoder (16) being adapted to compute said downmix signal (14) as a parametric representation of an intermediate signal of an M / S (middle / side) multi-channel audio signal; 2. The audio encoder of claim 1, wherein the multi-channel residual coder is configured to calculate a side signal corresponding to the intermediate signal of the M / S multi-channel audio signal, the multi-channel residual coder being configured to calculate a high band of the intermediate signal using a simulation of a time-domain bandwidth extension, or the multi-channel residual coder is configured to predict the high band of the intermediate signal using finding prediction information that minimizes a difference between a calculated side signal and a calculated full-band intermediate signal from a previous frame.

9. An audio decoder (102'') for decoding an encoded audio signal (103) comprising a core encoded signal, a bandwidth extension parameter and multi-channel information (20), said audio decoder (102'') comprising: a linear prediction domain core decoder (104) for decoding the core coded signal to generate a mono signal (142); an analysis filterbank (144) for converting said mono signal (142) into a spectral representation (145) partitioned into non-uniform parameter bands; a multi-channel decoder (146) for generating a first channel spectrum and a second channel spectrum from the spectral representation (145) of the mono signal (142) and the multi-channel information (20); a synthesis filter bank processor (148) for synthesis filtering the first channel spectrum to obtain a first channel signal, and for synthesis filtering the second channel spectrum to obtain a second channel signal; Including, The linear prediction domain core decoder comprises: a time-domain bandwidth extension processor (126) for generating a bandwidth-extended highband signal (140) from said bandwidth extension parameters and a lowband mono signal or said core coded signal, said bandwidth-extended highband signal (140) being a decoded highband signal of said audio signal; an ACELP decoder (120), a lowband synthesizer (122), and an upsampler (124) for outputting an upsampled lowband signal, the upsampled lowband signal being a decoded lowband mono signal; a combiner (128) configured to calculate a full-band ACELP decoded mono signal using the decoded low-band mono signal and the decoded high-band signal of the audio signal; a TCX decoder (130) and an Intelligent Gap Filling (IGF) processor (132) for obtaining a full-band TCX decoded mono signal; an all-band synthesis processor (134) for combining the full-band ACELP decoded mono signal and the full-band TCX decoded mono signal; The audio decoder (102'') as described above further comprises:

10. 10. The audio decoder (102'') of claim 9, wherein a crosspath (136) is provided for initializing the lowband synthesizer (122) with information derived by a lowband spectro-temporal transformation from the TCX decoder (130) and the intelligent gap filling IGF processor (132).

11. a frequency domain decoder (106); a second multi-channel decoder (110) for generating a second multi-channel representation (116) using an output of the frequency domain decoder (106) and second multi-channel information (22, 24); a first combiner (112) for combining the first channel signal and the second channel signal with the second multi-channel representation (116) to obtain a decoded audio signal (118); Further comprising: Audio decoder (102'') according to claim 9 or 10, wherein the second multi-channel decoder (110) is different from the multi-channel decoder (146).

12. 12. An audio decoder (102'') according to any of claims 9 to 11, wherein the analysis filterbank (144) includes a DFT for converting the mono signal (142) to the spectral representation (145) and the synthesis filterbank processor (148) includes an IDFT for converting the first channel spectrum to the first channel signal and the second channel spectrum to the second channel signal.

13. 13. The audio decoder of claim 12, wherein the analysis filter bank (144) is configured to apply a window to the DFT-transformed spectral representation (145) such that a right portion of the spectral representation of a previous frame and a left portion of the spectral representation of a current frame overlap, the previous frame and the current frame being consecutive.

14. An audio decoder (102'') for decoding an encoded audio signal (103) including a core encoded signal, bandwidth extension parameters, and multi-channel information (20), said audio decoder (102'') comprising: a linear prediction domain core decoder (104) for decoding the core coded signal to generate a mono signal (142); an analysis filterbank (144) for converting said mono signal (142) into a spectral representation (145) partitioned into non-uniform parameter bands; a multi-channel decoder (146) for generating a first channel spectrum and a second channel spectrum from the spectral representation (145) of the mono signal (142) and the multi-channel information (20); a synthesis filter bank processor (148) for synthesis filtering the first channel spectrum to obtain a first channel signal, and for synthesis filtering the second channel spectrum to obtain a second channel signal; Including, the mono signal (142) is an intermediate signal; The multi-channel decoder (146) obtains the first channel signal and the second channel signal from the mono signal (142), configured to obtain an M / S (middle / side) multi-channel decoded audio signal including a side signal, said side signal being calculated from said multi-channel information (20), configured to calculate an L / R multi-channel decoded audio signal for a low band using the multi-channel information (20) and the side signal, or an audio decoder (102'') configured to calculate a predicted side signal from the intermediate signal, and to calculate an L / R multi-channel decoded audio signal for a highband using the predicted side signal and an ILD (Inter-Channel Level Difference) value included in the multi-channel information (20).

15. The multi-channel decoder (146) performs complex operations on the L / R multi-channel decoded audio signal; Calculating the magnitude of the complex operation using the energy of the encoded intermediate signal and the energy of the L / R multi-channel decoded audio signal to obtain an energy compensation; 15. An audio decoder (102'') according to claim 14, configured to calculate the phase of the complex operation using an IPD (Inter-Channel Phase Difference) value contained in the multi-channel information (20).

16. A method (2000) for encoding a multi-channel signal (4), comprising the steps of: downmixing the multi-channel signal (4) to obtain a downmix signal (14), said downmixing comprising active downmixing or passive downmixing; - a linear predictive domain core coding step (16) of the downmix signal (14) to obtain an encoded downmix signal (26), the downmix signal (14) having a low band of the downmix signal (14) and a high band of the downmix signal (14), the linear predictive domain core coding step (16) of the downmix signal (14) comprising applying a bandwidth extension process to parametrically code the high band of the downmix signal (14); generating a spectral representation of said multi-channel signal (4); processing said spectral representation comprising a low band of said multi-channel signal (4) and a high band of said multi-channel signal (4) to generate multi-channel information (20); Including, The step of encoding the downmix signal (14) further comprises the step of decoding the encoded downmix signal (26) to obtain an encoded and decoded downmix signal (54), the method (2000) further comprises the step of calculating an encoded multi-channel residual signal (58) using the encoded and decoded downmix signal (54), the encoded multi-channel residual signal (58) representing an error between a decoded multi-channel representation obtained using the multi-channel information (20) and the multi-channel signal (4) before downmixing the multi-channel signal (4), the encoded downmix signal (14) being calculated as a multi-channel residual signal (58) representing an error between a decoded multi-channel representation obtained using the multi-channel information (20) and the multi-channel signal (4) before downmixing the multi-channel signal (4), the step of decoding the downmix signal (26) is configured to obtain, as the encoded and decoded downmix signal (54), only a low-band signal representing the low-band of the downmix signal (14), and the encoded multi-channel residual signal (58) has only a band corresponding to the low-band of the multi-channel signal (4) before downmixing the multi-channel signal (4), or the encoded multi-channel residual signal (58) has a low-band corresponding to the low-band of the multi-channel signal (4) before downmixing and a high-band corresponding to the high-band of the multi-channel signal (4) before downmixing, or The step of encoding the downmix signal (14) comprises a step of performing an ACELP process (30), the ACELP process being configured to operate on a downsampled downmix signal (34), the bandwidth extension process (36) being configured to parametrically encode the high band of the downmix signal (14) that has been removed from the downmix signal (14) by the downsampling, the step of encoding the downmix signal (14) comprises a TCX process (32), the TCX process (32) being configured to encode the high band of the downmix signal (14) by the downsampling, The method (2000) is configured to operate on the downmix signal (14) that has not been sampled or on the downmix signal (14) that has been downsampled to a lesser extent than the downsampling for the ACELP processing (30), and the TCX processing step includes a time-to-frequency transform step (40), a parameter generation step (42) for generating a parametric representation (46) of a first set of bands, and a quantization encoding step (44) for generating a set of quantized encoded spectral lines (48) for a second set of bands.

17. A method (2100) for decoding an encoded audio signal (103) including a core encoded signal, bandwidth extension parameters and multi-channel information (20), the method (2100) comprising: Linear prediction domain core decoding (104) the core coded signal to generate a mono signal (142); converting said mono signal (142) into a spectral representation (145) partitioned into non-uniform parameter bands; generating a first channel spectrum and a second channel spectrum from the spectral representation (145) of the mono signal (142) and the multi-channel information (20); synthesizing and filtering the first channel spectrum to obtain a first channel signal; and synthesizing and filtering the second channel spectrum to obtain a second channel signal; the mono signal (142) is an intermediate signal, and the step of generating the first channel spectrum and the second channel spectrum comprises the steps of obtaining an M / S multi-channel decoded audio signal including a side signal, the side signal being calculated from the multi-channel information (20), and calculating an L / R multi-channel decoded audio signal from the M / S multi-channel decoded audio signal for a low band using the multi-channel information (20) and the side signal, or calculating a predicted side signal from the intermediate signal and calculating the L / R multi-channel decoded audio signal for a high band using the predicted side signal and an ILD (Inter-Channel Level Difference) value included in the multi-channel information (20), or The step of decoding the core-coded signal includes a step of time-domain bandwidth extension processing (126) for generating a bandwidth-extended high-band signal (140) from the bandwidth extension parameters and a low-band mono signal or the core-coded signal, where the bandwidth-extended high-band signal (140) is the decoded high-band of the audio signal, a step of ACELP decoding (120), a step of low-band synthesis (122), and a step of upsampling to generate an upsampled low-band signal, which is the decoded low-band mono signal. A method (2100) comprising a processing step (124), a step of calculating a full-band ACELP decoded mono signal using the decoded low-band mono signal and the decoded high-band combination (128) of the audio signal, a step of TCX decoding (130) and intelligent gap filling processing (132) to obtain a full-band TCX decoded mono signal, and a full-band synthesis processing step (134) including a step of combining the full-band ACELP decoded mono signal and the full-band TCX decoded mono signal.

18. A computer program for performing the method according to claim 16 or 17 when the computer program runs on a computer or processor.

Images