BR122022012597B1 - AUDIO CODER AND DECODER USING A FREQUENCY DOMAIN PROCESSOR WITH FULL BAND GAP FILLING AND A TIME DOMAIN PROCESSOR - Google Patents

Abstract

An audio encoder for encoding an audio signal, comprising: a first encoding processor (600) of a first part of the audio signal in a frequency domain, characterized in that the first encoding processor (600) comprises: a time-frequency (602) for the first part of the audio signal in a frequency domain representation, having spectral lines up to a maximum frequency of the first part of the audio signal; an analyzer (604) of the frequency domain representation up to the maximum frequency to determine the primary spectral parts to be encoded with a first spectral resolution and the secondary spectral regions to be encoded with a second spectral resolution, the second spectral resolution being lower than the first spectral resolution; a spectral encoder (606) for encoding the primary spectral parts of the frequency domain representation with the first spectral resolution and for encoding the secondary spectral parts with the second spectral resolution; a second encoding processor (610) for a second part of the audio signal different in the time domain; a controller (620) configured for analyzing the audio signal and determining which part of the audio signal is the first encoded part of the audio signal (...).

Description

[0001] The present invention relates to encoding and decoding the audio signal and, in particular, processing the audio signal using parallel frequency domain and time domain encoder/decoder processors.

[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 lower bit rates have to be obtained, the coding employed leads to a reduction in audio quality which is often and mainly caused by a limitation on the encoder side of the bandwidth of the audio signal to be transmitted. Here, typically the audio signal is low-pass filtered so that no spectral waveform content remains above a certain predetermined cutoff frequency.

[0003] In modern codecs, there are well-known methods for restoring the signal on the decoder side through Bandwidth Extension (BWE) of the audio signal, e.g., Spectral Bandwidth Replication (SBR | Spectral Band Replication) that operates in the frequency domain or the so-called Time Domain Bandwidth Extension (TD-BWE), being a post-processor in speech encoders that operates in the time domain.

[0004] Additionally, several combined frequency domain/time domain coding concepts exist as concepts known under the term AMR-WB+ or USAC.

[0005] All of these combined time domain/coding concepts have in common that the frequency domain encoder relies on bandwidth extension technologies that incur a bandwidth limitation on the input audio signal and the part above a Crossing frequency or threshold frequency is encoded with a low-resolution coding concept and synthesized on the decoder side. Thus, such concepts mainly rely on a preprocessor technology on the encoder side and a corresponding post-processing functionality on the decoder side.

[0006] Typically, the time domain encoder is selected for useful signals to be decoded in the time domain, such as speech signal, and the frequency domain encoder is selected for non-speech signals, music signals, and etc. However, specifically for non-speech signals having prominent harmonics in the high frequency band, prior art frequency domain encoders have reduced accuracy and therefore reduced audio quality due to the fact that such prominent harmonics can only be parametrically encoded separately or are completely eliminated in the encoding/decoding process.

[0007] Furthermore, there are concepts in which the time domain encoding/decoding branch additionally depends on the extent of bandwidth that also parametrically encodes a higher frequency series, while a lower frequency series is typically encoded using an ACELP or any other type of CELP-related encoder, for example, a speech encoder. This bandwidth extension functionality increases bitrate efficiency, but on the other hand it still introduces inflexibility due to the fact that both coding branches, e.g. the frequency domain coding branch and the time domain coding branches are bandwidth limited due to the bandwidth extension procedure or the spectral band replication procedure operating above a certain crossover frequency substantially lower than the maximum frequency included in the input audio signal.

[0008] Relevant topics in the prior art comprise: - SBR as a post-processor for waveform decoding [1-3]; - USAC MPEG-D core switching [4]; - MPEG-H 3D IGF [5].

[0009] The following documents and patents describe the methods that are considered to constitute prior art for the application:

[0010] [1] M. Dietz, L. Liljeryd, K. Kjorling and O. Kunz, “Spectral Band Replication, a novel approach in audio coding”, at the 112th AES Convention, Munich, Germany, 2002.

[0011] [2] S. Meltzer, R. Bohm and F. Henn, “SBR enhanced audio codecs for digital broadcasting such as “Digital Radio Mondiale” (DRM)”, in 112th AES Convention, Munich, Germany, 2002.

[0012] [3] T. Ziegler, A. Ehret, P. Ekstrand and M. Lutzky, “Enhancing mp3 with SBR: Features and Capabilities of the new mp3PRO Algorithm”, at 112th AES Convention, Munich, Germany, 2002.

[0013] [4] USAC MPEG-D standard.

[0014] [5] PCT/EP2014/065109.

[0015] In USAC MPEG-D, a switchable central encoder is described. However, in USAC, the band-limited switch is restricted to always transmit a low-pass filtered signal. Therefore, certain music signals that contain prominent high-frequency content, for example, full-band sweeps, triangle sounds, etc., cannot be faithfully reproduced.

[0016] It is an object of the present invention to provide an improved concept for audio coding.

[0017] This object is achieved by an audio encoder according to claim 1, an audio decoder according to claim 11, an audio coding method according to claim 20, an audio decoding method according to claim audio according to claim 21 or a computer program according to claim 22.

[0018] The present invention is based on the discovery that a time domain encoding/decoding processor can be combined with a frequency domain encoding/decoding processor, having a gap-filling functionality, but this gap-filling functionality gap filling spectral holes is operated over the entire band of the audio signal or at least above a certain gap filling frequency. Essentially, the frequency domain encoding/decoding processor is particularly in a position to perform accurate encoding/decoding of either waveform or spectral value up to the maximum frequency and not just up to a cross frequency. Additionally, the full-bandwidth capability of the frequency domain encoder for high-resolution encoding allows integration of gap filling functionality into the frequency domain encoder.

[0019] Thus, according to the present invention, when using the full-band spectral encoder/decoder processor, problems related to the separation of bandwidth extension, on the one hand, and central coding, on the other, may be addressed and overcome by performing bandwidth extension in the same spectral domain in which the central decoder operates. Therefore, a full-rate core decoder is provided that encodes and decodes the full audio signal rate. This does not require the need for a sample rate reducer on the encoder side and a sample rate amplifier on the decoder side. Instead, all processing is performed in the full bandwidth or full sample rate domain. To obtain a high coding gain, the audio signal is analyzed to find a first set of spectral parts that have to be encoded with a high resolution, where this first set of primary spectral parts may include, in an application, tonal parts of the audio signal. On the other hand, non-tonal or noisy components in the audio signal constituting a second set of spectral parts are parametrically encoded with low spectral resolution. The encoded audio signal then requires only the first set of primary spectral parts encoded in a waveform-preserving manner with a high spectral resolution and, in addition, the second set of secondary spectral parts encoded parametrically at a low resolution using “portions”. ” of frequency coming from the first set. On the decoder side, the central decoder, which is a full-band decoder, reconstructs the first set of primary spectral parts in a waveform-preserving manner, that is, without any knowledge that there is any additional frequency regeneration. However, the spectrum thus generated has many spectral gaps. These gaps are subsequently filled with inventive Intelligent Gap Filling (IGF) technology by utilizing frequency regeneration applying parametric data on the one hand and utilizing a source spectral range, e.g., spectral parts reconstructed by the decoder. full rate audio on the other hand.

[0020] In other applications, the spectral parts, which are reconstructed only by noise filling rather than bandwidth replication or frequency portion filling, constitute a third set of third spectral parts. Due to the fact that the coding concept operates in a single domain for core coding/decoding on the one hand and frequency regeneration on the other hand, the IGF is not restricted to filling a higher frequency range, but can fill series frequency, by noise filling without frequency regeneration or by frequency regeneration using a frequency portion in a different frequency range.

[0021] Furthermore, it is emphasized that an information about spectral energies, an information about individual energies or an individual energy information, an information about a survival energy or a survival energy information, an information about a portion energy or a portion energy information, or a missing energy information or a missing energy information may comprise not only an energy value, but also an amplitude value (e.g. absolute), a level value or any another value, from which a final energy value can be derived. Thus, information about an energy may, for example, comprise the energy value itself, and/or a value of a level and/or an amplitude and/or an absolute amplitude.

[0022] An additional aspect is based on the observation that the correlation situation is not only important for the source range, but is also important for the target range. Furthermore, the present invention recognizes the situation that different correlation situations may occur in the source range and the target range. When, for example, a speech signal with high frequency noise is considered, the situation may be that the low frequency band comprising the speech signal with a small number of implications is highly correlated in the left channel and the right channel, when the speaker is placed in the middle. The high-frequency part, however, may be strongly uncorrelated due to the fact that there may be a different high-frequency noise on the left side compared to another high-frequency noise on the right side. Thus, when a direct-mode gap-filling operation is performed, which ignores this situation, then the high-frequency part would be correlated as well, and this can generate serious spatial segregation perturbations in the reconstructed signal. To resolve this issue, parametric data for a reconstruction band or, in general, for the second set of secondary spectral parts that have to be reconstructed using a first spectral set of primary spectral parts, are calculated to identify whether a first or a second representation of two different channels for the second spectral part or, indicated differently, for the reconstruction band. On the encoder side, a two-channel identification is therefore calculated for the secondary spectral parts, i.e. for the parts, for which additionally energy information for the reconstruction bands is calculated. A frequency regenerator on the decoder side then regenerates a second spectral part depending on a first part of the first set of primary spectral parts, for example, the source band and the parametric data for the second part, such as spectral envelope energy information. or any other spectral envelope data and additionally dependent on the identification of two channels for the second part, i.e. for that reconstruction band under reconsideration.

[0023] The two-channel identification is preferably transmitted as an identifier for each reconstruction band and this data is transmitted from an encoder to a decoder and the decoder then decodes the central signal as indicated by the identifiers preferably calculated for the central bands . So, in one implementation, the center signal is stored in both stereo representations (e.g., left/right and mid/side) and, for filling the IGF frequency portion, the source portion representation is chosen. to fit the target portion representation, as indicated, by the two-channel identification identifiers to the intelligent gap filling or reconstruction bands, i.e., to the target band.

[0024] It is emphasized that this procedure not only works for stereo signals, that is, for a left channel and for a right channel, but also operates for multiple channel signals. In the case of multi-channel signals, several different channel pairs can be processed as a right channel and a left channel as the first pair, a left edge channel and a right edge channel as the second pair, and a center channel and a LFE channel as the third pair. Other pairings can be determined for higher output channel formats, such as 7.1, 11.1 and others.

[0025] An additional aspect is based on the realization that the audio quality of the reconstructed signal can be improved through IGF, since the entire spectrum is accessible to the central encoder, so that, for example, perceptually important tonal parts in a high spectral range can still be encoded by the central encoder instead of parametric substitution. Additionally, a gap filling operation utilizing frequency portions from a first set of primary spectral parts which is, for example, a set of tonal parts typically from a lower frequency series, but also from a range higher frequency, if available, is performed. For decoder-side spectral envelope adjustment, however, the spectral parts from the first set of spectral parts located in the reconstruction band are no longer post-processed, for example by spectral envelope adjustment. Only the remaining spectral values in the reconstruction band that do not originate from the central decoder should be envelope fitted using envelope information. Preferably, the envelope information is full band envelope information accounting for the energy of the first set of primary spectral parts in the reconstruction band and the second set of secondary spectral parts in the same reconstruction band, where the last spectral values in the second set of Secondary spectral parts are indicated as zero and are therefore not encoded by the central encoder, but are parametrically encoded with low-resolution energy information.

[0026] It has been found that absolute energy values normalized or not with respect to the bandwidth of the corresponding band are useful and very efficient in an application on the decoder side. This applies especially when gain factors have to be calculated based on the residual energy in the reconstruction band, the missing energy in the reconstruction band and the frequency portion information in the reconstruction band.

[0027] Furthermore, it is preferred that the encoded bitstream not only covers power information for the reconstruction bands, but additionally scale factors for scale factor bands up to the maximum frequency. This ensures that for each reconstruction band, for which a certain tonal part, that is, a first spectral part, is available, that first set of the first spectral part can actually be decoded with a right amplitude. Furthermore, in addition to the scaling factor for each reconstruction band, an energy for that reconstruction band is generated in an encoder and transmitted to a decoder. Furthermore, it is preferable for reconstruction bands to coincide with scale factor bands or, in the case of energy pooling, at least reconstruction band boundaries to coincide with scale factor band boundaries.

[0028] An additional aspect is based on the finding that certain deficiencies in audio quality can be remedied by applying an adaptive signal frequency portion filling scheme. To this end, an encoder-side analysis is performed to find the best corresponding source region candidate for a given target region. A corresponding information that identifies a target region from a given source region along with optionally some additional information is generated and transmitted as additional information to the decoder. The decoder then applies a frequency chunk filling operation using the corresponding information. For this purpose, the decoder reads the corresponding information from the transmitted data stream or data file and accesses the source region identified for a certain reconstruction band and, if indicated in the corresponding information, additionally performs some processing of this source region data to generate the raw spectral data for the reconstruction band. Then, this result of the frequency portion filling operation, i.e., the raw spectral data for the reconstruction band, is modeled using the spectral envelope information in order to finally obtain a reconstruction band comprising the primary spectral parts as tonal parts too. These tonal parts, however, are not generated by the adaptive portion filling scheme, but these primary spectral parts are output by the audio decoder or central decoder directly.

[0029] The adaptive spectral portion selection scheme can operate with a low granularity. In this implementation, a source region is subdivided into typically overlapping source regions and the target region or reconstruction bands are given by the non-overlapping frequency target regions. Then, the similarities between each source region and each target region are determined on the encoder side, and the best matching pair of a source region and the target region is identified by the corresponding information, and on the decoder side, the identified source region in the corresponding information is used to generate the raw spectral data for the reconstruction band.

[0030] In order to obtain a higher granularity, each source region is allowed to alternate, in order to obtain a certain delay where similarities are maximum. This delay can be as fine as a frequency position and allows for even better matching between a source region and target region.

[0031] Furthermore, in addition to just identifying a best identification pair, this correlation delay can also be transmitted within the corresponding information and, additionally, even a signal can be transmitted. When the signal is determined to be negative on the encoder side, then a corresponding signal is also transmitted within the corresponding information, and on the decoder side, the spectral values of the source region are multiplied by “-1” or, in a complex representation , are “rotated” 180 degrees.

[0032] A further implementation of this invention is applied to a portion bleaching operation. Whitening a spectrum removes the coarse spectral envelope information and emphasizes the spectral fine structure that is of primary interest for evaluating portion similarity. Therefore, a portion of the frequency on the one hand and/or the source signal on the other sides are whitened before calculating a cross-correlation measure. When only the portion is whitened using a predefined procedure, a whitening identifier is transmitted indicating to the decoder that the same predefined whitening process should be applied to the frequency portion within the IGF.

[0033] Regarding portion selection, it is preferred to use the correlation identifier to spectrally exchange the regenerated spectrum by an integer transform unit. Depending on the underlying transform, spectral switching may require the addition of corrections. In the case of extraneous delays, the portion is additionally modulated by multiplying by an alternating temporal sequence of -1/1 to compensate for the inverted frequency representation of all other bands within the MDCT. Furthermore, the sign of the correlation result is applied when generating the frequency portion.

[0034] Furthermore, it is preferable to use portion removal and stabilization to make sure that disturbances created by rapidly changing the source regions to the same reconstruction region or target region are avoided. For this purpose, a similarity analysis between different identified source regions is performed and when a source portion is similar to other source portions with a similarity above a threshold, then that source portion can be discarded from the set of source portions. potential origin since it is highly correlated with other portions of origin. Furthermore, as a type of portion selection stabilization, it is preferred to maintain the portion order from the previous frame if none of the source portions in the current frame correlate (better than a given threshold) with the target portions in the current frame. current frame.

[0035] A further aspect is based on the finding that improved quality and reduced bit rate, specifically for signals comprising transient parts as they occur very frequently in audio signals, is obtained by combining Temporal Noise Modulation (TNS) technology | Temporal Noise Shaping) or Temporal Portion Modulation (TTS | Temporal Tile Shaping) with high-frequency reconstruction. TNS/TTS processing on the encoder side being implemented by a frequency prediction reconstructs the time envelope of the audio signal. Depending on the implementation, for example, when the temporal noise modulation filter is determined within a frequency range not only covering the source frequency range, but also the target frequency range to be reconstructed in a frequency regeneration decoder , the temporal envelope is not only applied to the central audio signal up to an initial gap-filling frequency, but the temporal envelope is also applied to the spectral bands of reconstructed secondary spectral parts. Thus, pre-echoes and post-echoes that would occur without temporal portion modulation are reduced or eliminated. This is achieved by applying an inverse prediction on the frequency not only within the center frequency range up to a certain gap-filling start frequency, but also within a frequency range above the center frequency range. For this purpose, frequency regeneration or frequency portion generation is performed on the decoder side before applying a prediction on the frequency. However, the frequency prediction can be applied before or after the spectral envelope modulation depending on whether the energy information calculation was performed on the residual spectral values subsequent to filtering or on the (full) spectral values before the envelope modulation.

[0036] TTS processing over one or more frequency portions additionally establishes a correlation continuity between the source band and the reconstruction rate or across two adjacent reconstruction rates or frequency portions.

[0037] In one implementation, it is preferred to use TNS/TTS complex filtering. This way, bandwidth distortion (temporal) perturbations of a critically sampled real representation, such as MDCT, are avoided. A TNS complex filter can be calculated on the encoder side by applying not only a modified discrete cosine transform, but also a modified discrete sine transform in addition to obtaining a modified complex transform. Regardless, only the modified discrete cosine transform values, for example the real part of the complex transform, are transmitted. On the decoder side, however, it is possible to estimate the imaginary part of the transform using the MDCT spectrum of previous or subsequent frames, so that on the decoder side the complex filter can again be applied in inverse prediction over frequency, and specifically , the prediction about the boundary between the source range and the reconstruction rate and also about the boundary between adjacent frequency portions within the reconstruction rate.

[0038] The inventive audio coding system efficiently encodes arbitrary audio signals over a wide range of bit rates. While, for high bit rates, the inventive system converges towards transparency, for low bit rates the noticeable annoyance is minimized. Therefore, the main part of the available bit rate is used to encode waveforms only for the most perceptually relevant structures of the signal in the encoder, and the resulting spectral gaps are filled in the decoder with signal content that closely approximates the original spectrum. . A very limited amount of bit is consumed to control the driven parameter, also called Intelligent Gap Filling (IGF) by the dedicated side information transmitted from the encoder to the decoder.

[0039] In other applications, the time domain encoding/decoding processor relies on a lower sampling rate and corresponding bandwidth extension functionality.

[0040] In additional applications, a cross-processor is provided to initialize the time domain encoder/decoder with initialization data derived from the currently processed frequency domain encoder/decoder signal. This allows that when the currently processed part of the audio signal is processed by the frequency domain encoder, the parallel time domain encoder is initialized so that when a switch from the frequency domain encoder to a time domain encoder happens, this time domain encoder can start processing since all the initialization data related to the previous signals is already there due to the cross processor. The cross-processor is preferably applied on the encoder side, additionally on the decoder side and preferably utilizes a frequency-time transform which additionally performs a very efficient sampling rate reduction from the higher output or input sampling rate. to the lower time domain central encoder sampling rate by selecting only certain low band part of the domain signal along with certain reduced transform size. Thus, a sample rate conversion from high sample rate to low sample rate is very efficiently performed and this signal obtained by the transform with the reduced transform size can then be used to initialize the mode time domain encoder/decoder. that the time domain encoder/decoder is ready to immediately perform time domain encoding when this situation is signaled by the controller and the immediately preceding part of the audio signal has been encoded in the frequency domain.

[0041] Thus, preferred applications of the present invention enable integrated switching of a perceptual audio encoder comprising spectral gap filling and a time domain encoder with or without bandwidth extension.

[0042] Thus, the present invention relies on methods that are not restricted to removing high frequency content above a cutoff frequency in the frequency domain encoder of the audio signal, but adaptively removes the signal spectral regions of bandpass leaving spectral gaps in the encoder and subsequently reconstructs these spectral gaps in the decoder. Preferably, an integrated solution such as smart gap filling is used that efficiently combines full bandwidth audio coding and spectral gap filling particularly in the MDCT transform domain.

[0043] Thus, the present invention provides an improved concept for combining speech coding and a subsequent time domain bandwidth extension with a full-band waveform decoding comprising spectral gap filling in an encoder/ switchable perceptual decoder.

[0044] Thus, in contrast to existing methods, the new concept uses the coding of the waveform of the full-band audio signal in the transform domain encoder and at the same time enables integrated switching with a speech encoder preferably followed by an extension of the time domain bandwidth.

[0045] Additional applications of the present invention prevent the explained problems from occurring due to fixed bandwidth limitation. The concept enables the switchable combination of a full-bank waveform encoder in the frequency domain equipped with a spectral gap filler and a lower sample rate speech encoder and an extension of the time domain bandwidth. Such an encoder is capable of encoding the shape of the aforementioned problematic signals by providing full audio bandwidth up to the Nyquist frequency of the input audio signal. Nevertheless, seamless switching between both encoding strategies is guaranteed particularly by cross-processor applications. For this integrated switching, the cross-processor represents a cross-connection in both the encoder and the decoder between the full-rate encoder capable of full-band (input sampling rate) frequency domain and the low-rate ACELP encoder having a lower sampling rate to correctly initialize the ACELP parameters and buffers particularly within the adaptive codebook, the LPC filter, or the resampling stage, in switching from the frequency domain encoder such as TCX to the time domain encoder such as ACELP.

[0046] The present invention is subsequently discussed in relation to the accompanying drawings, in which:

[0047] Figure 1a illustrates an apparatus for encoding an audio signal;

[0048] Figure 1b illustrates a decoder for decoding a coded audio signal that combines with the encoder of figure 1a;

[0049] Figure 2a illustrates a preferred implementation of the decoder;

[0050] Figure 2b illustrates a preferred implementation of the encoder;

[0051] Figure 3a illustrates a schematic representation of a spectrum, as generated by the spectral domain decoder of Figure 1b;

[0052] Figure 3b illustrates a table, indicating the relationship between the scale factors for the scale factor bands and energies for the reconstruction bands and noise filling information for a noise filling band;

[0053] Figure 4a illustrates the functionality of the spectral domain encoder for applying the selection of spectral parts to the first and second sets of spectral parts;

[0054] Figure 4b illustrates an implementation of the functionality of figure 4a;

[0055] Figure 5a illustrates a functionality of an MDCT encoder;

[0056] Figure 5b illustrates a decoder functionality with MDCT technology;

[0057] Figure 5c illustrates an implementation of the frequency regenerator;

[0058] Figure 6 illustrates an implementation of an audio encoder;

[0059] Figure 7a illustrates a cross processor within the audio encoder;

[0060] Figure 7b illustrates an implementation of an inverse or time-frequency transform that additionally provides a sampling rate reduction within the cross-processor;

[0061] Figure 8 illustrates a preferred implementation of the controller of Figure 6;

[0062] Figure 9 illustrates an additional application of the time domain encoder having bandwidth extension functionalities;

[0063] Figure 10 illustrates a preferred use of a preprocessor;

[0064] Figure 11a illustrates a schematic implementation of the audio decoder;

[0065] Figure 11b illustrates a cross-processor within the decoder for providing initialization data to the time domain decoder;

[0066] Figure 12 illustrates a preferred implementation of the time domain decoding processor of Figure 11a;

[0067] Figure 13 illustrates a further implementation of time domain bandwidth extension;

[0068] Figure 14a illustrates a preferred implementation of an audio encoder;

[0069] Figure 14b illustrates a preferred implementation of an audio decoder;

[0070] Figure 14c illustrates an inventive implementation of a time domain decoder with sample rate conversion and bandwidth extension.

[0071] Figure 6 illustrates an audio encoder for encoding an audio signal, comprising a first coding processor 600 for encoding a first part of the audio signal in a frequency domain. The first coding processor 600 comprises a time-frequency converter 602 for converting the first part of the audio input signal into a frequency domain representation having spectral lines up to a maximum frequency of the input signal. Furthermore, the first coding processor 600 comprises an analyzer 604 for analyzing the frequency domain representation up to the maximum frequency to determine the primary spectral regions to be encoded with a first spectral representation and to determine the secondary spectral regions to be encoded with a second spectral resolution being lower than the first spectral resolution. In particular, the full-band analyzer 604 determines which frequency lines or spectral values in the spectrum of the time-frequency converter are to be spectrally line encoded and which other parts are to be encoded in a parametric form, and these later spectral values are then , reconstructed on the decoder side with the gap filling procedure. The actual coding operation is performed by a spectral encoder 606 for encoding the primary spectral regions or spectral parts with the first resolution and for parametrically coding the secondary spectral regions or parts with the second spectral resolution.

[0072] The audio encoder of Figure 6 additionally comprises a second coding processor 610 for encoding part of the audio signal in a time domain. Additionally, the audio encoder comprises a controller 620 configured for analyzing the audio signal at an input audio signal 601 and determining which part of the audio signal is the first encoded part of the audio signal in the frequency domain and which part of the audio signal is the second encoded part of the audio signal in the time domain. Furthermore, a coded signal modulator 630 which may be, for example, implemented as a bit stream multiplexer is provided that is configured for modulating a coded audio signal comprising a first coded part of the signal to the first part of the audio signal. audio and a second part of the signal encoded for the second part of the audio signal. Importantly, the encoded signal has only a frequency domain representation or a time domain representation of one and the same part of the audio signal.

[0073] Thus, controller 620 ensures that, for a single portion of the audio signal, only one time domain representation or one frequency domain representation is in the encoded signal. This can be accomplished by controller 620 in several ways. One way would be that for one and the same part of the audio signal, both representations arrive at block 630 and the controller 620 controls the coded signal modulator 630 to introduce only one of both representations to the coded signal. Alternatively, however, controller 620 may control an input to the first encoding processor and an input to the second encoding processor such that, based on analysis of the corresponding portion of the signal, only one of either block 600 or 610 is activated to actually perform the complete encoding operation and the other block is deactivated.

[0074] This deactivation may be a deactivation or, as illustrated with respect to, for example, Figure 7a, it is just a type of “initialization” mode where the other encoding processor is only active to receive and process the initialization data in order to initialize the internal memories, but any specific encoding operation is not performed in any way. This activation can be done by a certain switch on the input that is not illustrated in Figure 6 or, preferably, by control lines 621 and 622. Thus, in this application, the second encoding processor 610 does not output anything when the controller 620 has determined that part of the current audio signal should be encoded by the first encoding processor, but the second encoding processor is, however, provided with the initialization data to be active for immediate switching in the future. On the other hand, the first encoding processor is configured not to need any past data to update any internal memories, and therefore, when the portion of the current audio signal is to be encoded by the second encoding processor 610, then the controller 620 may control the first final coding processor 600 via control line 621 to be completely inactivated. This means that the first encoding processor 600 does not need to be in an initialization state or standby state, but may be in a complete shutdown state. This is particularly preferable for mobile devices where power consumption and therefore battery life is an issue.

[0075] In the additional specific implementation of the second coding processor operating in the time domain, the second coding processor comprises a sample rate reducer 900 or sample rate converter for converting the portion of the audio signal into a representation with a lower sampling rate, wherein the lower sampling rate is less than a sampling rate at the input to the first encoding processor. This is illustrated in Figure 9. In particular, when the audio input signal comprises a low band and a high band, it is preferred that the lower sample rate representation at the output of block 900 has only the low band of the audio portion. audio input signal and this low band is then encoded by a time domain low band encoder 910 that is configured for time domain encoding of the lower sample rate representation provided by block 900. Furthermore, the Time domain bandwidth extension encoder 920 is provided to parametrically encode the high bandwidth. To this end, the time domain bandwidth extension encoder 920 receives at least the high band of the audio input signal or the low band and the high band of the audio input signal.

[0076] In a further application of the present invention, the audio encoder further comprises, although not illustrated in figure 6, but illustrated in figure 10, a preprocessor 1000 configured for preprocessing the first part of the audio signal and of the second part of the audio signal. In one application, this preprocessor comprises a prediction analyzer for determining prediction coefficients. This prediction analyzer can be implemented as a linear prediction coding (LPC) analyzer to determine LPC coefficients. However, other analyzers can also be implemented. Furthermore, the preprocessor, also illustrated in Figure 14a, comprises a prediction coefficient quantizer 1010, characterized by the device illustrated in Figure 14a receiving prediction coefficient data from the prediction analyzer, also illustrated in Figure 14a, in 1002.

[0077] Furthermore, the preprocessor additionally comprises an entropy encoder for generating an encoded version of the quantized prediction coefficients. It is important to note that the coded signal modulator 630 or the specific implementation, i.e., the bitstream multiplexer 613 ensures that the coded version of the quantized prediction coefficients is included in the coded audio signal 632. Preferably, the LPC coefficients are not directly quantized, but are converted into an ISF, for example, or whatever representation is best suited for quantization. This conversion is preferably performed by determining block 1002 of LPC coefficients or is performed within block 1010 for quantization of LPC coefficients.

[0078] Additionally, the preprocessor may comprise a resampler 1004 for resampling an input audio signal at an input sampling rate to a lower sampling rate for the time domain encoder. When the time domain encoder is an ACELP encoder having a certain ACELP sampling rate, then the reduction of the sampling rate is performed to preferably 12.8 kHz or 16 kHz. The input sample rate can be any of a specific number of sample rates such as 32 kHz or even a higher sample rate. On the other hand, the sampling rate of the time domain encoder will be predetermined by certain constraints and the resampler 1004 performs this resampling and outputs the lower sampling rate representation of the input signal. Thus, the resampler 1004 may perform similar functionality and may further be one or the same element as the sample rate reducer 900 illustrated in the context of FIG. 9.

[0079] Furthermore, it is preferable to apply a pre-emphasis in the pre-emphasis block 1005 in Figure 14a. Pre-emphasis processing is well known in the time domain coding technique and is described in the literature with reference to AMR-WB+ processing and pre-emphasis is particularly configured for the compensation of a spectral slope and therefore enables a better calculation of LPC parameters in a given LPC order.

[0080] Furthermore, the preprocessor may additionally comprise an extraction of the TCX-LTP parameter to control an LTP post-filter at 1420 in Figure 14b. This block is illustrated at 1006 in figure 14a. Furthermore, the preprocessor may additionally comprise other features illustrated at 1007 and such other features may comprise a tone search feature, a voice activity detection (VAD) feature, or any other features known in the art. timing or speech coding.

[0081] As illustrated, the result of block 1006 is inserted into the coded signal, that is, in the application of figure 14a, inserted into the bit stream multiplexer 630. Furthermore, if necessary, the data from block 1007 can also be introduced into the bitstream multiplexer or may alternatively be used for time domain coding purposes in the time domain encoder.

[0082] Thus, to summarize, common to both passes is a preprocessing operation 1000 in which generally used signal processing operations are performed. These comprise a resampling to a sample rate of ACELP (12.8 or 16 kHz) for a parallel pass and this resampling is always performed. Furthermore, an extraction of the TCX-LTP parameter illustrated in block 1006 is performed and, additionally, a pre-emphasis and a determination of LPC coefficients are performed. As described, pre-emphasis compensates for spectral skew and therefore makes the calculation of LPC parameters at a given LPC order more efficient.

[0083] Subsequently, reference is made to Figure 8 in order to illustrate a preferred implementation of the controller 620. The controller receives, at an input, the part of the audio signal under consideration. Preferably, as illustrated in Figure 14a, the controller receives any signal available in preprocessor 1000 which may be the original input signal at the input sample rate or a resampled version at the lower time domain encoder sample rate or a signal obtained subsequent to pre-emphasis preprocessing in block 1005.

[0084] Based on this portion of the audio signal, controller 620 directs a frequency domain encoder simulator 621 and a time domain encoder simulator 622 to calculate for each encoder the possibility of an estimated signal ratio for the noise. Subsequently, selector 623 selects the encoder that provided the best signal-to-noise ratio, naturally in consideration of a predefined bit rate. The selector then identifies the corresponding encoder through the control output. When it is determined that the portion of the audio signal under consideration is to be encoded using the frequency domain encoder, the time domain encoder is set to an initialization state or in other applications that do not require very immediate switching into a state. completely disabled. However, when it is determined that the portion of the audio signal under consideration should be encoded by the time domain encoder, the frequency domain encoder is then disabled.

[0085] Subsequently, a preferred implementation of the controller illustrated in Figure 8 is illustrated. The decision whether to choose the ACELP or TCX pass is made in the switching decision by simulating the ACELP and TCX encoder and switching to the best branch in realization. For this, the SNR of the ACELP and TCX branch is estimated based on a simulation of the ACELP and TCX encoder/decoder. TCX encoder/decoder simulation is performed without TNS/TTS analysis, IGF encoder, quantization/arithmetic circuit encoder, or without any TCX decoder. Instead, the TCX SNR is estimated using an estimate of the quantizer distortion in the modulated MDCT domain. The ACELP encoder/decoder simulation is performed using only an adaptive codebook simulation and innovative codebook. The ACELP SNR is simply estimated by computing the distortion introduced by an LTP filter in the weighted signal domain (adaptive codebook) and scaling the distortion by a constant factor (innovative codebook). Thus, complexity is greatly reduced compared to an approach where TCX and ACELP coding is performed in parallel. The branch with the highest SNR is chosen for subsequent full encoding execution.

[0086] In the case where the TCX branch is chosen, a TCX decoder is executed in each structure that outputs a signal at the ACELP sampling rate. This is used to update the memories used for the ACELP coding pass (residual LPC, Mem w0, memory de-emphasis), to allow immediate switching from TCX to ACELP. The memory update is performed on each TCX pass.

[0087] Alternatively, a complete analysis by synthesis process can be performed, i.e., both encoder simulators 621, 622 implement the actual encoding operations and the results are compared by selector 623. Alternatively, again, A complete feed forward calculation can be done by performing a signal analysis. For example, when the signal is determined to be a speech signal by a signal classifier, the time domain encoder is selected and when the signal is determined to be a music signal, then the frequency domain encoder is selected. selected. Other procedures in order to distinguish between both encoders based on a signal analysis of the part of the audio signal under consideration can also be applied.

[0088] Preferably, the audio encoder additionally comprises a crossover processor 700 illustrated in figure 7a. When the frequency domain encoder 600 is active, the crossover processor 700 provides initialization data to the time domain encoder 610 so that the time domain encoder is ready for an integrated switch on a portion of the future signal. In other words, when the portion of the current signal is determined to be encoded using the frequency domain encoder, and when it is determined by the controller that the immediately following portion of the audio signal is to be encoded by the time domain encoder 610 then , without the cross-processor, such an out-of-the-box integrated switch would not be possible. The cross-processor, however, provides a signal derived from the frequency domain encoder 600 to the time domain encoder 610 for the purpose of initializing the memories in the time domain encoder since the time domain encoder 610 has a dependency on a current structure of the input or encoded signal of a structure immediately preceding it in time.

[0089] Thus, the time domain encoder 610 is configured to be initialized by the initialization data in order to encode a portion of the audio signal following an encoded portion of the previous audio signal by the frequency domain encoder 600 efficiently. .

[0090] In particular, the cross-processor comprises a time converter for converting a frequency domain representation into a time domain representation that may be forwarded to the time domain encoder directly or after some additional processing. This converter is illustrated in figure 14a as an IMDCT (inverse modified discrete cosine transform | inverse modified discrete cosine transform) block. This block 702, however, has a different transform size compared to the time-frequency converter block 602 indicated in the block of Figure 14a (modified discrete cosine transform block). As indicated in block 602, the time-frequency converter 602 operates at the input sampling rate and the inverse modified discrete cosine transform 702 operates at the lower ACELP sampling rate.

[0091] The ratio of the time domain encoder sampling rate or ACELP sampling rate and the frequency domain encoder sampling rate or input sampling rate can be calculated and be a reduction factor for the DS sampling rate illustrated in Figure 7b. Block 602 has a large transform size and IMDCT block 702 has a small transform size. As illustrated in Figure 7b, the IMDCT block 702 therefore comprises a selector 726 for selecting the lower spectral part of an input to the IMDCT block 702. The full-band spectrum part is defined by the sample rate reduction factor. DS. For example, when the lower sample rate is 16 kHz and the input sample rate is 32 kHz, then the sample rate reduction factor is 0.5 and therefore selector 726 selects the lower half of the sample rate spectrum. full band. When the spectrum has, for example, 1024 MDCT lines, then the selector selects the bottom 512 MDCT lines.

[0092] This low-frequency portion of the full-band spectrum is inserted into a small-size, unfoldable-block transform 720, as illustrated in Figure 7b. The transform size is also selected according to the sampling rate reduction factor and is 50% of the transform size in block 602. A synthesis windowing with a window with a small number of coefficients is then performed. The number of synthesis window coefficients is equal to the sampling rate reduction factor multiplied by the number of analysis window coefficients used by block 602. Finally, an overlap addition operation is performed with a small number of operations per block and the number of operations per block is again the number of operations per block in a full-rate implementation MDCT multiplied by the sample rate reduction factor.

[0093] Thus, a very efficient sampling rate reduction operation can be applied since sampling rate reduction is included in the IMDCT implementation. In this context, it is emphasized that block 702 may be implemented by an IMDCT, but may also be implemented by any other transform or filter bank implementation that can appropriately scale into the core of the actual transform and other transform-related operations.

[0094] In an additional application illustrated in Figure 14a, the time frequency converter comprises additional functionalities in addition to the analyzer. The analyzer 604 of Figure 6 may comprise in the application of Figure 14a a temporal noise modulation/time portion modulation analysis block 604a that operates as discussed in the context of Figure 2b, block 222, for the TNS/TTS analysis block. 604a and illustrated with respect to Figure 2b for the tonal mask 226 that corresponds to the IGF encoder 604b in Figure 14a.

[0095] Furthermore, the frequency domain encoder preferably comprises a noise modulation block 606a. The noise modulation block 606a is controlled by the quantized LPC coefficients as generated by the block 1010. The quantized LPC coefficients used for noise modulation 606a perform a spectral modulation of the high-resolution spectral values or directly encoded spectral lines (instead of parametrically coded) and the result of block 606a is similar to the spectrum of a signal subsequent to an LPC filtering stage operating in the time domain as an LPC analysis filtering block 704 to be described later. Furthermore, the output of noise modulation block 606a is then quantized and entropy encoded as indicated by block 606b. The result of block 606b corresponds to the first encoded portion of the audio signal or an encoded portion of the frequency domain audio signal (along with additional information).

[0096] The crossover processor 700 comprises a spectral decoder for calculating a decoded version of the first encoded portion of the signal. In the application of Figure 14a, the spectral decoder 701 comprises an inverse noise modulation block 703, a gap filling decoder 704, a TNS/TTS synthesis block 705, and the IMDCT block 702 discussed previously. These blocks undo the specific operations performed by blocks 602 to 606b. In particular, a noise modulation block 703 undoes the noise modulation performed by block 606a based on quantized LPC coefficients 1010. IGF decoder 704 operates as discussed with respect to Figure 2A, blocks 202 and 206, and block TNS/TTS synthesis block 705 operates as discussed in the context of block 210 of Figure 2A and the spectral decoder further comprises the IMDCT block 702. Furthermore, the cross processor 700 in Figure 14a additionally or alternatively comprises a delay stage 707 for inputting a delayed version of the decoded version obtained by the spectral decoder 701 into a de-emphasis stage 617 of the second encoding processor for the purpose of initializing the de-emphasis stage 617.

[0097] Furthermore, the cross-processor 17 may additionally or alternatively comprise a weighted prediction coefficient analysis filtering stage 708 to filter the decoded version and to input a filtered decoded version to a codebook determiner. 613 indicated as “MMSE” in figure 14a of the second encoding processor to initialize this block. Additionally or alternatively, the cross-processor comprises the LPC analysis filtering stage for filtering the decoded version of the first encoded portion of the signal emitted by the spectral decoder 700 to an adaptive codebook stage 712 for initializing the block 612. Additionally or alternatively, the cross-processor also comprises a pre-emphasis stage 709 for performing pre-emphasis preprocessing on the decoded version emitted by a spectral decoder 701 prior to LPC filtering. The output pre-emphasis stage may also be inserted into an additional delay stage 710 for the purposes of initializing an LPC synthesis filtering block 616 within the time domain encoder 610 for the purposes of initializing this analysis filtering block. from LPC 611.

[0098] The processor time domain encoder 610 comprises, as illustrated in Figure 14a, a pre-emphasis that operates at the lower ACELP sampling rate. As illustrated, this pre-emphasis is the pre-emphasis performed in the pre-processing stage 1000 and has the reference number 1005. The pre-emphasis data is entered into an LPC analysis filtering stage 611 that operates in the domain time and this filter is controlled by the quantized LPC coefficients 1010 obtained by the preprocessing stage 1000. As known from AMR-WB+ or USAC encoders or other CELP encoders, the residual signal generated by block 611 is fed to a book adaptive codebook 612 and further, the adaptive codebook 612 is connected to an innovative codebook stage 614 and the codebook data of the adaptive codebook 612 and the innovative codebook are input into the stream multiplexer of bits as illustrated.

14a. This block cooperates with the innovative codebook block 614. Furthermore, the time domain encoder further comprises a decoder part having an LPC synthesis filtering block 616, a de-emphasis block 617 and a adaptive bass post-filter stage 618 for calculating the parameters for an adaptive bass post-filter which is, however, applied on the decoder side. Without any adaptive low post-filtering on the decoder side, blocks 616, 617, 618 would not be necessary for the time domain encoder 610.

[0100] As illustrated, several blocks of the time domain decoder depend on the prior signals and these blocks are the adaptive codebook block, the codebook determiner 613, the LPC synthesis filtering block 616 and the of de-emphasis 617. These blocks are provided with the cross-processor data derived from the frequency domain encoding processor data in order to initialize these blocks for the purpose of being ready for an instantaneous switch from the frequency domain encoder to the time domain. As can also be seen from figure 14a, any dependence on prior data is not necessary for the frequency domain encoder. Therefore, the cross-processor 700 does not provide any initialization data from the time domain encoder memory to the frequency domain encoder. However, for other implementations of the frequency domain encoder, where past dependencies exist and where memory initialization data is required, the cross processor 700 is configured to operate in both directions.

[0101] A preferred application of an audio encoder therefore comprises the following parts:

[0102] The preferred audio decoder is described below: The waveform decoder portion consists of a full-band TCX decoder pass with IGF both operating at the codec input sample rate. In parallel, an alternative pass of the ACELP decoder at the lower sampling rate exists which is further enhanced downstream by a TD-BWE.

[0103] For ACELP initialization when switching from TCX to ACELP, a crossover (consisting of a split TCX decoder frontend, but additionally providing output at the lower sample rate and some post-processing) exists that performs ACELP initialization. Inventive ACELP. Sharing the same sample rate and filter order between TCX and ACELP on LPCs allows for easier and more efficient ACELP initialization.

[0104] To visualize the switching, two switches are drawn in 14b. While the second downstream switch chooses between the TCX/IGF or ACELP/TD-BWE output, the first switch pre-updates the buffers in the resampling QMF stage downstream of the ACELP pass through the cross-pass output or simply passes the by ACELP.

[0105] Subsequently, audio decoder implementations in accordance with aspects of the present invention are discussed in the context of figures 11a to 14c.

[0106] An audio decoder for decoding a coded audio signal 1101 comprises a first decoding processor 1120 for decoding a first coded part of the audio signal in a frequency domain. The first decoding processor 1120 comprises a spectral decoder 1122 for decoding primary spectral regions with a high spectral resolution and for synthesizing the secondary spectral regions using a parametric representation of the secondary spectral regions and at least one first decoded spectral region to obtain a spectral representation decoded. The decoded spectral representation is a full-band decoded spectral representation as discussed in the context of Figure 6 and as also discussed in the context of Figure 1a. Generally speaking, the first decoding processor therefore comprises a full-band implementation with a gap filling procedure in the frequency domain. The first decoding processor 1120 further comprises a time-frequency converter 1124 for converting the decoded spectral representation into a time domain to obtain a first decoded portion of the audio signal.

[0107] Furthermore, the audio decoder comprises a second decoding processor 1140 for decoding the second encoded portion of the audio signal in the time domain to obtain a second decoded portion of the signal. Furthermore, the audio decoder comprises a combiner 1160 for combining the first decoded part of the signal and the second decoded part of the signal to obtain a decoded audio signal. The decoded portions of the signal are combined in the sequence that is also illustrated in Fig. 14b by an implementation of the switch 1160 that represents an implementation of the combiner 1160 of Fig. 11a.

[0108] Preferably, the second decoding processor 1140 is a time domain bandwidth extending processor and comprises, as illustrated in Figure 12, a time domain low bandwidth decoder 1200 for decoding a time domain signal. of low band time. This implementation further comprises a sample rate amplifier 1210 for increasing the sampling rate of the low band time domain signal. Additionally, a time domain bandwidth extension decoder 1220 is provided for synthesizing a high band of the output audio signal. Furthermore, a mixer 1230 is provided for mixing a synthesized high-band of the time domain output signal and an amplified sampling rate low-band time domain signal to obtain the output of the time domain encoder. Thus, block 1140 in Figure 11a can be implemented by the functionality of Figure 12 in a preferred application.

[0109] Figure 13 illustrates a preferred application of the time domain bandwidth extension decoder 1220 of Figure 12. Preferably, a time domain sampling rate amplifier 1221 is provided receiving, as an input, a signal LPC residual of a time domain low-band decoder included within block 1140 and illustrated at 1200 in Figure 12 and further illustrated in the context of Figure 14b. The time domain sample rate amplifier 1221 generates a sample rate amplified version of the residual LPC signal. This version is then inserted into a non-linear distortion block 1222 which generates, based on its input signal, an output signal having higher frequency values. A nonlinear distortion can be a copy, a reflection, a frequency shift, or a nonlinear device such as a diode or transistor operated in the nonlinear region. The output signal from block 1222 is input to an LPC synthesis filter block 1223 that is controlled by the LPC data used for the low-band decoder as well as specific envelope data generated by the LPC bandwidth extension block. time domain 920 on the encoder side of Figure 14a, for example. The output of the LPC synthesis block is then fed into a bandpass or high pass filter 1224 to finally obtain the high band, which is then fed into the mixer 1230, as illustrated in figure 12.

[0110] Subsequently, a preferred implementation of the sample rate amplifier 1210 of Figure 12 is discussed in the context of Figure 14b. The sample rate amplifier preferably comprises an analysis filter bank that operates at a first sample rate of the time domain low-band decoder. A specific implementation of such an analysis filter bank is a QMF analysis filter bank 1471 illustrated in Figure 14b. Additionally, the sample rate amplifier comprises a synthesis filter bank 1473 that operates at a second output sample rate being higher than the first time domain low-band sample rate. Thus, the QMF synthesis filter bank 1473 which is a preferred implementation of the general filter bank operates at the output sample rate. When the DS sample rate reduction factor, as discussed in the context of Figure 7b, is 5, then the QMF analysis filter bank 1471 has, for example, only 32 filter bank channels and the QMF analysis filter bank 1471 has, for example, only 32 filter bank channels and the QMF synthesis 1473 has, for example, 64 QMF channels, but the upper half of the filter bank channels, i.e. the upper 32 filter bank channels are inserted with zeros or noise, while the upper 32 filter bank channels lower filters are inserted with the corresponding signals provided by the QMF analysis filter bank 1471. Preferably, however, a bandpass filtering 1472 is performed within the QMF filter bank domain in order to ensure that the output QMF 1473 synthesis is a version with the amplified sample rate of the ACELP decoder output, but without any disturbances above the maximum frequency of the ACELP decoder.

[0111] Other processing operations may be performed within the QMF domain in addition to, or instead of, bandpass filtering 1472. If no processing is performed in any way, then QMF analysis and QMF synthesis constitute a efficient sample rate amplifier 1210.

[0112] Subsequently, the construction of the individual elements in figure 14b are discussed in more detail.

[0113] The full-band frequency domain decoder 1120 comprises a first decoding block 1122a for decoding the high-resolution spectral coefficients and for additionally performing noise filling in the low-band part as known, for example, from USAC technology. Furthermore, the full band decoder comprises an IGF 1122b processor for filling the spectral holes using synthesized spectral values that have been encoded only parametrically and therefore encoded with a low resolution on the encoder side. Then, in block 1122c, an inverse noise modulation is performed and the result is fed into a TNS/TTS synthesis block 705 which provides, as a final output, an input into a time-frequency converter 1124, which is preferably implemented as an inverse modified discrete cosine transform that operates on the output, i.e. high sampling rate.

[0114] Additionally, a harmonic post-filter or LTP is used and is controlled by data obtained by the TCX-LTP parameter extraction block 1006 in figure 14b. The result is then the first decoded part of the audio signal at the output sampling rate and as can be seen from figure 14b, this data has the high sampling rate and therefore any additional frequency boosting is of no consequence. 1a to 5c.

[0115] Various elements in Figure 14b are very similar to the corresponding blocks in the crossover processor 700 of Figure 14a, particularly with respect to the IGF decoder 704 corresponding to the IGF processing 1122b and the inverse noise modulation operation controlled by the quantized LPC coefficients. 1145 corresponds to the inverse noise modulation 703 of Figure 14a and the TNS/TTS synthesis block 705 in Figure 14b corresponds to the synthesis of the TNS/TTS block 705 in Figure 14a. Importantly, however, the IMDCT block 1124 in Fig. 14b operates at the high sampling rate while the IMDCT block 702 in Fig. 14a operates at a low sampling rate. Thus, the block 1124 in Figure 14b comprises the large-size transform and unfolding block 710, the synthesis window in block 712, and the overlap addition stage 714 with the corresponding large number of operations, large number of window coefficients, and a large size of the transform compared to the corresponding resources 720, 722, 724, which are operated in block 702 and, as will be described later, in block 1171 of the cross processor 1170 in Figure 14b as well.

[0116] The time domain decoding processor 1140 preferably comprises the time domain or ACELP low-band decoder 1200 comprising an ACELP decoder stage 1149 for obtaining the decoded gains and novel codebook information. Additionally, an ACELP codebook adaptive stage 1141 is provided and a subsequent ACELP post-processing stage 1142 and a final synthesis filter as LPC synthesis filter 1143, which is again controlled by the obtained quantized LPC coefficients 1145 of the bitstream demultiplexer 1100 corresponding to the encoded signal analyzer 1100 in Figure 11a. The output of the LPC synthesis filter 1143 is input to a de-emphasis stage 1144 to cancel or nullify the processing introduced by the pre-emphasis stage 1005 of the preprocessor 1000 of Figure 14a. The result is the time domain output signal at a low sampling rate and a low bandwidth and in the case where frequency domain output is required, the switch 1480 is in the indicated position and the output of the de- emphasis 1144 is introduced to the sample rate amplifier 1210 and then mixed with the high bands of the time domain bandwidth extension decoder 1220.

[0117] In accordance with the applications of the present invention, the audio decoder additionally comprises the crossover processor 1170 illustrated in figure 11b and figure 14b for calculating, from the decoded spectral representation of the first encoded part of the audio signal, of the initialization data of the second decoding processor so that the second decoding processor is initialized to decode the second encoded portion of the audio signal following in time the first portion of the audio signal in the encoded audio signal, i.e., so that the time domain decoding processor 1140 is ready for instantaneous switching from one part of the audio signal to the next without any loss in quality or efficiency.

[0118] Preferably, the crossover processor 1170 comprises an additional time-frequency converter 1171 that operates at a lower sampling rate than the time-frequency converter of the first decoding processor in order to obtain a first decoded portion of the additional signal. in the time domain to be used as the initialization signal or for which any initialization data can be derived. Preferably, this IMDCT or low sampling rate time-frequency converter is implemented as illustrated in Figure 7b, item 726 (selector), item 720 (small size and unfoldable transform), synthesis windowing with a small number of window coefficients as indicated at 722 and an overlap addition stage with a small number of operations as indicated at 724. Thus, the IMDCT block 1124 in the frequency domain full-band decoder is implemented as indicated by block 710, 712, 714, and the IMDCT block 1171 is implemented as indicated in Figure 7b by block 726, 720, 722, 724. Again, the sample rate reduction factor is the ratio of the sample rate of the field domain encoder to time or the low sampling rate and the highest frequency domain sampling rate or output sampling rate and this sampling rate reduction factor is less than 1 and can be any number greater than 0 and less than 1.

[0119] As illustrated in Figure 14b, the crossover processor 1170 further comprises, alone or with other elements, a delay stage 1172 for delaying the first decoded portion of the additional signal and for inserting the first decoded delayed portion of the signal into a stage de-emphasis 1144 of the second decoding processor for initialization. Furthermore, the crossover processor additionally or alternatively comprises a pre-emphasis filter 1173 and a delay stage 1175 for filtering and delaying the first decoded portion of the additional signal and for providing the delayed output of the block 1175 in a stage LPC 1143 synthesis filtering of the ACELP decoder for initialization purposes.

[0120] Furthermore, the cross-processor may comprise, alternatively or in addition to the other elements mentioned, an LPC analysis filter 1174 for generating a predictive residual signal of the first decoded portion of the additional signal or a pre-emphasized first decoded portion. of the additional signal and to input the data to a codebook synthesizer of the second decoding processor and, preferably, to the codebook adaptive stage 1141. Furthermore, the output of the time-frequency converter 1171 with the low rate Sampling is also input to the QMF analysis stage 1471 of the sample rate amplifier 1210 for initialization purposes, i.e., when the currently decoded portion of the audio signal is delivered by the frequency domain full-band decoder 1120.

[0121] The preferred audio decoder is described below: The waveform decoder portion consists of a full-band TCX decoder pass with IGF that operates at the input sample rate of the codec. In parallel, an alternative pass of the ACELP decoder at the lower sampling rate exists and is further enhanced downstream by a TD-BWE.

[0122] For ACELP initialization when switching from TCX to ACELP, a crossover (which consists of a split TCX decoder frontend, but additionally provides output at the lower sample rate and some post-processing) exists performing ACELP initialization inventive. Sharing the same sample rate and filter order between TCX and ACELP on LPCs enables easier and more efficient ACELP initialization.

[0123] To visualize the switching, two switches are drawn in figure 14b. While the second downstream switch chooses between TCX/IGF or ACELP/TD-BWE output, the first switch pre-updates the QMF buffers downstream of the ACELP pass through the cross-pass output or simply passes the ACELP.

[0124] To summarize, the preferred aspects of the invention that can be used alone or in combination relate to a combination of an ACELP and TD-BWE encoder with a full-band capable TCX/IGF technology preferably associated using a crossover signal .

[0125] An additional specific feature is a crossover signal pass for ACELP initialization to enable integrated switching.

[0126] An additional aspect is that a short IMDCT is inserted with a bottom of long high-rate MDCT coefficients to efficiently implement a cross-pass sample rate conversion.

[0127] An additional feature is an efficient realization of partially split cross-pass with a full-band TCX/IGF in the decoder.

[0128] An additional feature is the crossover signal pass for QMF initialization to allow integrated switching from TCX to ACELP.

[0129] An additional feature is a crossover signal pass to QMF enabling compensation of the delay gap between the resampled ACELP output and a TCX/IGF output of the filter bank when switching from ACELP to TCX.

[0130] An additional aspect is that an LPC is provided for both the TCX and ACELP encoders at the same sampling rate and filter order, although the TCX/IGF encoder/decoder is full band capable.

[0131] Subsequently, Figure 14c is discussed as a preferred implementation of a time domain decoder that operates as a standalone decoder or in combination with the full band capable frequency domain decoder.

[0132] Generally, the time domain decoder comprises an ACELP decoder, a subsequently connected resampler or sample rate amplifier, and a time domain bandwidth extension functionality. Particularly, the ACELP decoder comprises an ACELP decoding stage for recovering gains and the innovative codebook 1149, an ACELP codebook adaptive stage 1141, an ACELP postprocessor 1142, an LPC synthesis filter 1143 controlled by the quantized LPC coefficients of a bitstream demultiplexer or encoded signal analyzer and the subsequently connected de-emphasis stage 1144. Preferably, the residual time domain signal, being at a sampling rate of ACELP, is input in a time domain bandwidth extension decoder 1220, which provides high bandwidth at the outputs.

[0133] In order to amplify the sampling rate of the de-emphasis output 1144, a sampling rate amplifier comprising the QMF analysis block 1471 and the QMF synthesis block 1473 are provided. Within the filter bank domain defined by blocks 1471 and 1473, a bandpass filter is preferably applied. In particular, as previously discussed, the same features can also be used that were discussed with respect to the same reference numbers. Furthermore, the time domain bandwidth extension decoder 1220 may be implemented as illustrated in FIG. 13 and generally comprises increasing the sampling rate of the residual ACELP signal or residual time domain signal in the ACELP sampling rate finally to a bandwidth extended signal output sampling rate.

[0134] Subsequently, additional details regarding the frequency domain encoder and decoder being full band capable are discussed with respect to Figures 1A to 5C.

[0135] Figure 1a illustrates an apparatus for encoding an audio signal 99. The audio signal 99 is input into a time spectrum converter 100 for converting an audio signal having a sampling rate into a spectral representation 101 emitted by the time spectrum converter. The spectrum 101 is input into a spectral analyzer 102 for analysis of the spectral representation 101. The spectral analyzer 101 is configured to determine a first set of primary spectral parts 103 to be encoded with a first spectral resolution and a second different set of spectral parts secondary 105 to be encoded with a second spectral resolution. The second spectral resolution is lower than the first spectral resolution. The second set of secondary spectral parts 105 is input into a parameter calculator or parametric encoder 104 for calculating spectral envelope information having the second spectral resolution. Furthermore, a spectral domain audio encoder 106 is provided for generating a first encoded representation 107 of the first set of primary spectral parts having the first spectral resolution. Furthermore, the parameter calculator/parametric encoder 104 is configured to generate a second encoded representation 109 of the second set of secondary spectral parts. The first coded representation 107 and the second coded representation 109 are input into a bitstream multiplexer or bitstream modulator 108 and the block 108 finally outputs the coded audio signal for transmission or storage in a storage device.

[0136] Typically, a first spectral part such as 306 of figure 3a will be surrounded by two secondary spectral parts such as 307a, 307b. This is not the case in HE-AAC, where the frequency range of the central encoder is band limited.

[0137] Figure 1b illustrates a decoder that matches the encoder of figure 1a. The first encoded representation 107 is input to an audio spectral domain decoder 112 for generating a first decoded representation of a first set of primary spectral parts, the decoded representation having a first spectral resolution. Furthermore, the second encoded representation 109 is input to a parametric decoder 114 for generating a second decoded representation of a second set of secondary spectral parts having a second spectral resolution being lower than the first spectral resolution.

[0138] The decoder further comprises a frequency regenerator 116 for regenerating a second reconstructed spectral part, having the first spectral resolution using a first spectral part. The frequency regenerator 116 performs a portion filling operation, that is, it uses a portion or part of the first set of primary spectral parts and copies this first set of primary spectral parts into the reconstruction band or into the reconstruction band having the second part spectral and typically performs spectral envelope modulation or other operation as indicated by the second decoded representation output by parametric decoder 114, i.e., using information about the second set of secondary spectral parts. The first set of decoded primary spectral parts and the second set of reconstructed spectral parts, as indicated in the output of the frequency regenerator 116 at line 117, are input into a spectrum-time converter 118 configured for conversion of the first decoded representation and the second spectral part reconstructed into a time representation 119, the time representation having a certain high sampling rate.

[0139] Figure 2b illustrates an implementation of the encoder of Figure 1a. An input audio signal 99 is input into an analysis filter bank 220 corresponding to the time spectrum converter 100 of Figure 1a. Then, a temporal noise modulation operation is performed in the TNS block 222. Therefore, the input to the spectral analyzer 102 of Fig. temporal noise/time portion modulation is not applied or may be spectral residual values, when the TNS operation as illustrated in Figure 2b, block 222 is applied. For two-channel signals or multi-channel signal, a splice channel coding 228 can additionally be performed, so that the spectral domain encoder 106 of Figure 1a can comprise the splice channel coding block 228. Furthermore, an entropy encoder 232 for performing lossless data compression is provided and is also a part of the spectral domain encoder 106 of Figure 1a.

[0140] The spectral analyzer/tonal mask 226 separates the TNS output of the block 222 into the central band and the tonal components corresponding to the first set of primary spectral parts 103 and the residual components corresponding to the second set of secondary spectral parts 105 of Figure 1a . The block 224 indicated as the IGF parameter extraction coding corresponds to the parametric encoder 104 of Figure 1a and the bitstream multiplexer 230 corresponds to the bitstream multiplexer 108 of Figure 1a.

[0141] Preferably, the analysis filter bank 222 is implemented as an MDCT (modified discrete cosine transform filter bank) and the MDCT is used to transform the signal 99 into a time-frequency domain with the cosine transform modified discrete module that acts as the frequency analysis tool.

[0142] The spectral analyzer 226 preferably applies a tone mask. This tone mask estimation stage is used to separate the tonal components from the noise-like components in the signal. This enables the central encoder 228 to encode all tonal components with a physical-acoustic module. The pitch mask estimation stage can be implemented in several different forms and is preferably similar in functionality to the sine range estimation stage used in sine and noise modeling for speech/audio coding [8, 9] or an audio encoder based on the HILN model described in [10]. Preferably, an implementation is used that is easy to implement without the need to maintain start-stop trajectories, but any other tone or noise detector can be used as well.

[0143] The IGF module calculates the similarity that exists between a source region and a target region. The target region will be represented by the spectrum from the source region. The measurement of similarity between target and source regions is performed using a cross-relationship approach. The target region is divided into non-overlapping frequency portions. For each portion in the target region, source portions are created from a fixed initial frequency. These source portions overlap by a factor between 0 and 1, where 0 means 0% overlap and 1 means 100% overlap. Each of these source portions is correlated with the target portion at various delays to find the source portion that best matches the target portion. The best matching portion number is stored in , the delay at which it best correlates with the target is stored in , and the correlation signal is stored in ^•.l!^N'o;iicasto.^emmiir]líi!aí ^ . . . In the case where the correlation is highly negative, the source portion needs to be multiplied by -1 before the portion filling process in the decoder. The IGF module also takes care not to overwrite the tonal components in the spectrum as the tonal components are preserved using the tone mask. One energy parameter per band is used to store the energy of the target region allowing us to reconstruct the spectrum precisely.

[0144] This method has certain advantages over classical SBR [1] in that the harmonic grid of a multitonal signal is preserved by the central encoder while only the gaps between the sinusoids are filled with the best corresponding “modulated noise” from the source region. . Another advantage of this system compared to ASR (Accurate Spectral Replacement) [2-4] is the absence of a signal synthesis stage that creates important parts of the signal in the decoder. Instead, this task is taken over by the central encoder, allowing the preservation of important components of the spectrum. Another advantage of the proposed system is the continuous scalability that the resources offer. Just using and , for each portion is called coarse-grained matching and can be used for low bit rates while using the variable for each portion allows us to match the target and source spectra.

[0145] Furthermore, a pick stabilization technique is proposed by removing frequency domain disturbances such as vibration and musical noise.

[0146] In the case of stereo channel pairs, additional stereo junction processing is applied. This is necessary because over a certain target range the signal can be a highly dispersed and correlated sound source. In case the source regions chosen for that particular region are not well correlated, although the energies are matched for the target regions, the spatial image may suffer due to the uncorrelated source regions. The encoder analyzes each energy band of the target region, typically cross-correlating the spectral values and if a certain threshold is exceeded, sets a junction indicator for that energy band. In the decoder, the right and left channel power bands are treated individually if this stereo junction indicator is not set. In case junction stereo indicator is set, both power and repair are performed in the joint stereo domain. The junction stereo information for the IGF regions is signaled similar to the junction stereo information for the core coding, including an indicator indicating in case of prediction the direction of the prediction to be from downmix to residual or vice versa.

[0147] The energies can be calculated from the energies transmitted in the L/R domain.

[0148] being the frequency index in the transform domain.

[0149] Another solution is to calculate and transmit the energies directly in the joint stereo domain for bands where joint stereo is active, so no additional energy transformation is required on the decoder side.

[0150] The source portions are always created according to the Median/Lateral Matrix:

[0151] Power adjustment:

[0152] Stereo connection -> LR transformation:

[0153] If no additional prediction parameters are encoded:

[0154] If no additional prediction parameters are encoded and if the signaled direction is from middle to side:

[0155] If the direction indicated is from the side to the middle:

[0156] This processing ensures that from the portions used for regeneration of highly correlated target regions and displaced target regions, the resulting left and right channels still represent a correlated and displaced sound source even if the source regions are uncorrelated. preserving the stereo image for such regions.

[0157] In other words, in the bitstream, stereo set indicators are transmitted indicating whether L/R or M/S as an example for overall stereo set coding should be used. In the decoder, first, the center signal is decoded, as indicated by the stereo set indicators for the center bands. Second, the central signal is stored in the representation of both L/R and M/S. For filling the IGF portion, the representation of the source portion is chosen to fit the representation of the target portion as indicated by the stereo junction information for the IGF bands.

[0158] Temporal Noise Modulation (TNS) is a standard technique and part of AAC [11 - 13]. TNS can be considered as an extension of the basic scheme of a perceptual encoder, inserting an optional processing step between the filter bank and the quantization stage. The main task of the TNS module is to hide the quantization noise produced in the temporal masking region from similar transient signals and thus leads to a more efficient coding scheme. First, TNS calculates a set of prediction coefficients using “forward prediction” in the transform domain, for example, MDCT. These coefficients are then used to flatten the temporal envelope of the signal. Because quantization affects the TNS-filtered spectrum, the quantization noise is temporarily flattened. By applying inverse TNS filtering on the decoder side, the quantization noise is modulated according to the temporal envelope of the TNS filter, and therefore the quantization noise is marked by the transient.

[0159] IGF is based on a representation of MDCT. For efficient encoding, preferably long blocks of approximately 20 ms should be used. If the signal within such a long block contains transients, audible pre- and post-echoes occur in the IGF spectral bands due to the filling of the portion. Figure 7c shows a typical pre-echo effect before the transient onset due to IGF. On the left side, the spectrogram of the original signal is shown and on the right side the spectrogram of the bandwidth extended signal without TNS filtering is shown.

[0160] This pre-echo effect is reduced using TNS in the context of IGF. Here, TNS is used as a time portion modulation (TTS) tool, as spectral regeneration in the decoder is performed on the residual TNS signal. The necessary prediction TTS coefficients are calculated and applied using the full spectrum on the encoder side as usual. The TNS/TTS start and stop frequencies are not affected by the IGF start frequency of the IGF tool. Compared to legacy TNS, the stop frequency of TTS is increased to the stop frequency of the IGF tool, which is higher than . On the decoder side, the TNS/TTS coefficients are applied to the complete spectrum again, i.e., the central spectrum plus the referred spectrum plus the tonal components of the tone map (see figure 7e). The application of TTS is necessary to form the temporal envelope of the said spectrum to match the envelope of the original signal again. Thus the pre-echoes shown are reduced. Furthermore, it still modulates the quantization noise in the signal below as usual with TNS.

[0161] In legacy decoders, spectral repair in an audio signal corrupts the spectral correlation at the repair edges and thus impairs the temporal envelope of the audio signal by introducing scatter. Thus, another benefit of filling the IGF portion in the residual signal is that, after applying the modulation filter, the edges of the portion are continuously correlated, resulting in a more faithful temporal reproduction of the signal.

[0162] In an inventive encoder, the spectrum having passed through TNS/TTS filtering, tone mask processing and IGF parameter estimation is devoid of any signal above the initial IGF frequency except for the tonal components. This sparse spectrum is now encoded by the central encoder using arithmetic and predictive coding principles. These encoded components together with the signaling bits form the audio bit stream.

[0163] Figure 2a illustrates the corresponding implementation of the decoder. The bitstream in Figure 2a corresponding to the encoded audio signal is input into the demultiplexer/decoder which would be connected, with respect to Figure 1b, to blocks 112 and 114. The bitstream demultiplexer separates the audio input signal into the first coded representation 107 of figure 1b and in the second coded representation 109 of figure 1b. The first coded representation having the first set of primary spectral parts is inserted into the joint channel decoding block 204 corresponding to the spectral domain decoder 112 of Figure 1b. The second coded representation is inserted into the parametric decoder 114 not illustrated in Figure 2a and then inserted into the IGF block 202 corresponding to the frequency regenerator 116 of Figure 1b. The first set of primary spectral parts required for frequency regeneration is inserted into the IGF block 202 via line 203. Furthermore, subsequent to decoding the joint channel 204 specific core decoding is applied to the tonal mask block 206 so that the output of the tonal mask 206 corresponds to the output of the spectral domain decoder 112. Then, a combination by the combiner 208 is performed, i.e., a structure construction where the output of the combiner 208 now has the full range spectrum, but still in the domain filtered by TNS/TTS. Then, in block 210, an inverse TNS/TTS operation is performed using the TNS/TTS filter information provided via line 109, i.e., the TTS side information is preferably included in the first encoded representation generated by the domain encoder. spectral 106 which may, for example, be a central forward encoder of AAC or USAC, or may also be included in the second encoded representation. At the output of block 210, a full spectrum up to maximum frequency is provided with the full range frequency being defined by the sampling rate of the original input signal. Then, a spectrum/time conversion is performed in the synthesis filter bank 212 to finally obtain the output audio signal.

[0164] Figure 3a illustrates a schematic representation of the spectrum. The spectrum is subdivided into SCB scale factor bands where there are seven scale factor bands SCB1 to SCB7 in the example illustrated in figure 3a. The scale factor bands can be AAC scale factor bands that are defined in the AAC standard and have increasing bandwidth at higher frequencies, as illustrated in Figure 3a schematically. It is preferred to perform smart gap filling at the first point in the spectrum, i.e., at low frequencies, but to start the IGF operation at an IGF starting frequency illustrated in 309. Therefore, the center frequency band extends from the frequency lowest for the initial IGF frequency. Above the IGF starting frequency, spectrum analysis is applied to separate the high-resolution spectral components 304, 305, 306, 307 (the first set of primary spectral parts) from the low-resolution components represented by the second set of secondary spectral parts. . Figure 3a illustrates a spectrum that is exemplarily input to the spectral domain encoder 106 or the joint channel encoder 228, i.e., the central encoder operates in the full range, but encodes a significant amount of zero spectral values. that is, these zero spectral values are quantized to zero or are set to zero prior to quantization or subsequent to quantization. In any case, the central encoder operates in full range, i.e., as if the spectrum were as illustrated, i.e., the central decoder must not necessarily be aware of any intelligent gap filling or encoding of the second set of secondary spectral parts with lower spectral resolution.

[0165] Preferably, the high resolution is defined by an in-line encoding of spectral lines as MDCT lines, while the second resolution or low resolution is defined, for example, by calculating only a single spectral value per band from the fact that scale, where a scale factor band spans several frequency lines. Thus, the second low resolution is, with respect to its spectral resolution, much smaller than the first or high resolution defined by the in-line coding typically applied by the central encoder such as an AAC or USAC central encoder.

[0166] With reference to the scale factor or energy calculation, the situation is illustrated in figure 3b. Because the encoder is a central encoder and because there may, but does not necessarily have to, be components of the first set of spectral parts in each band, the central encoder calculates a scale factor for each band not only in the central band below the starting frequency of IGF 309, but also above the starting frequency of IGF up to the maximum frequency that is small or equal to half the sampling frequency, i.e. fs/2. Thus, the tonal coded parts 302, 304, 305, 306, 307 of figure 3a and in this application together with the scale factors SCB1 to SCB7 correspond to high-resolution spectral data. The low-resolution spectral data is calculated starting from the IGF starting frequency and corresponds to the energy information values E1, E2, E3, E4, which are transmitted together with the scaling factors SF4 to SF7.

[0167] Particularly, when the central encoder is under a low bit rate condition, an additional noise filling operation in the center band, i.e., lower in frequency than the IGF initial frequency, i.e., in bands of the scaling factor SCB1 to SCB3 can be further applied. In noise filling, there are several adjacent spectral lines that have been quantized to zero. On the decoder side, these zero-quantized spectral values are resynthesized and the resynthesized spectral values are adjusted in magnitude using a noise filling energy such as NF2 illustrated at 308 in Figure 3b. The noise filling energy, which can be given in absolute terms or in relative terms with respect to the scale factor as in USAC corresponds to the energy of the set of spectral values quantized to zero. These noise filling spectral lines can also be considered as a third set of tertiary spectral parts that are referred to by direct noise filling synthesis without any IGF operation depending on frequency regeneration utilizing the frequency portions of other frequencies for reconstruction of the portions of the frequency using spectral values from a source band and the energy information E1, E2, E3, E4.

[0168] Preferably, the bands for which the energy information is calculated coincide with the scale factor bands. In other applications, a grouping of the energy information value is applied so that, for example, for scale factor bands 4 and 5, only a single energy information value is transmitted, but still in this application, the edges of the pooled reconstruction bands coincide with the edges of the scale factor bands. If different bandwidth operations are applied, then certain recalculations or synchronization calculations may be applied, and this may make sense depending on a certain implementation.

[0169] Preferably, the spectral domain encoder 106 of Figure 1a is a psychoacoustically driven encoder, as illustrated in Figure 4a. Typically, as, for example, illustrated in the MPEG2/4 AAC or MPEG1/2 Layer 3 standard, the audio signal to be encoded after being transformed into a spectral range (401 in figure 4a) is routed to a calculator scale factor 400. The scale factor calculator is controlled by a psychoacoustic model that additionally receives the audio signal to be quantized or that receives, as in Layer 3 of MPEG1/2 or AAC standard for MPEG, a complex spectral representation of the audio signal. The psychoacoustic model calculates, for each scale factor band, a scale factor that represents the psychoacoustic limit. Additionally, the scaling factors are then, by cooperation of well-known internal and external interaction circuits or by any suitable coding procedure adjusted so that certain bit rate conditions are met. Then, the spectral values to be quantized on the one hand and the calculated scale factors on the other hand are input into a quantizing processor 404. In direct operation of the audio encoder, the spectral values to be quantized are weighted by the scale factors and, the weighted spectral values are then input into a fixed quantizer typically having compression functionality in higher amplitude ranges. So, at the output of the quantizer processor there are quantization indices that are then forwarded into an entropy encoder typically having specific and very efficient coding for a set of zero quantization indices for adjacent frequency values or, as also called in technique, a “run” of zero values.

[0170] In the audio encoder of Figure 1a, however, the quantizer processor typically receives information about the secondary spectral parts from the spectral analyzer. Thus, the quantizer processor 404 makes sure that, at the output of the quantizer processor 404, the secondary spectral parts as identified by the spectral analyzer 102 are zero or have a representation recognized by an encoder or a decoder as a representation of zero that can be very efficiently encoded, specifically when there are “runs” of zero values in the spectrum.

[0171] Figure 4b illustrates an implementation of the quantizer processor. The MDCT spectral values can be entered into a zero setting in block 410. Then, the secondary spectral parts are already set to zero before a weighting by the scale factors in block 412 is performed. In a further implementation, block 410 is not provided, but the zeroing operation is performed in the block 418 subsequent to the weighting block 412. In yet another additional implementation, the zeroing operation may also be performed in a block 410. zero setting 422 subsequent to a quantization in quantizer block 420. In this implementation, blocks 410 and 418 would not be present. Generally, at least one of blocks 410, 418, 422 is provided depending on the specific implementation.

[0172] Then, at the output of block 422, a quantized spectrum is obtained corresponding to what is illustrated in figure 3a. This quantized spectrum is then input to an entropy encoder such as 232 in figure 2b which can be a Huffman encoder or an arithmetic encoder as, for example, defined in the USAC standard.

[0173] The zeroing blocks 410, 418, 422, which are provided alternatively to each other or in parallel are controlled by the spectral analyzer 424. The spectral analyzer preferably comprises any implementation of a well-known hue detector or comprises any type other than operative detector for separating a spectrum into components to be encoded with a high resolution and components to be encoded with a low resolution. Other such algorithms implemented in the spectral analyzer may be a voice activity detector, a noise detector, a speech detector or any other detector that decides, depending on the spectral information or metadata associated with the resolution requirements for different spectral parts.

[0174] Figure 5a illustrates a preferred implementation of the time spectrum converter 100 of Figure 1a as, for example, implemented in AAC or USAC. The time spectrum converter 100 comprises a windower 502 controlled by a transient detector 504. When the transient detector 504 detects a transient, then a transition from long windows to short windows is signaled to the windower. The windower 502 then calculates, for overlay blocks, the window structures, where each window structure typically has two N values such as 2048 values. Then, a transformation within a block transformer 506 is performed, and this block transformer typically and additionally provides a decimation, so that a combined decimation/transform is performed to obtain a spectral structure with N values as MDCT spectral values. Thus, for a long window operation, the structure at the input of block 506 comprises two N values as 2048 values and a spectral structure then has 1024 values. Then, however, a switch is performed in short blocks, when eight short blocks are performed, where each short block has 1/8 windowed time domain values compared to a long window and each spectral block has 1/8 spectral values compared to a long block. Thus, when this decimation is combined with 50% windower overlap operation, the spectrum is a critically sampled version of the 99 audio time domain signal.

[0175] Subsequently, reference is made to Figure 5b which illustrates a specific implementation of the frequency regenerator 116 and the spectrum-time converter 118 of Figure 1b, or the combined operation of blocks 208, 212 of Figure 2a. In Figure 5b, a specific reconstruction band is considered as scale factor 6 band in Figure 3a. The first spectral part in this reconstruction band, i.e., the first spectral part 306 of Figure 3a is inserted into the structure creator/regulator block 510. Furthermore, a second reconstructed spectral part for the scale factor 6 band is also , inserted into structure creator/regulator 510. Furthermore, energy information such as E3 of Figure 3b for a band of scale factor 6 is also inserted into block 510. The second reconstructed spectral part in the reconstruction band has already been generated by filling a frequency portion using a source band and the reconstruction band then corresponds to the target band. Now, an energy adjustment of the structure is performed to finally obtain the complete reconstructed structure having the N values as, for example, obtained in the output of combiner 208 in figure 2a. Then, in block 512, an inverse transform/interpolation of the block is performed to obtain 248 time domain values for, for example, the 124 spectral values at the input of block 512. Then, a synthesis windowing operation is performed in the block 514 which is again controlled by a long window/short window indication transmitted in the encoded audio signal. Then, in block 516, an overlap/add operation with a preperiod is performed. Preferably, MDCT applies 50% overlap so that for each new period of 2 N, N values, time domain values are finally output. A 50% overlap is heavily preferred due to the fact that it provides critical sampling and crossover from one structure to the next structure due to the overlap/add operation in block 516.

[0176] As illustrated at 301 in Figure 3a, a noise filling operation can additionally be applied not only below the IGF starting frequency, but also above the IGF starting frequency as for the complete reconstruction band that coincides with the band of scale factor 6 in figure 3a. Then, the noise fill spectral values can also be entered into the structure creator/regulator 510 and the adjustment of the noise fill spectral values can also be applied within this block or the noise fill spectral values can already be adjusted using the noise filling energy before it is fed into the 510 structure creator/regulator.

[0177] Preferably, an IGF operation, that is, a frequency portion filling operation using spectral values from other parts can be applied to the complete spectrum. Thus, a spectral portion filling operation can not only be applied in the high band above an initial IGF frequency, but can also be applied in the low band. Furthermore, noise filling without frequency portion filling can also be applied not only below the IGF starting frequency, but also above the IGF starting frequency. However, it has been found that high-quality and high-efficiency audio coding can be obtained when the noise filling operation is limited to the frequency range below the IGF starting frequency and when the frequency portion filling operation is restricted to the frequency range above the IGF initial frequency, as illustrated in figure 3a.

[0178] Preferably, the target portions (TT | target tiles) (having frequencies greater than the IGF starting frequency) are linked to the edges of the scale factor band of the full rate encoder. The source portions (ST | source tiles), from which the information is taken, i.e. for frequencies lower than the IGF starting frequency are not bound by the edges of the scale factor band. The size of the ST should correspond to the size of the associated TT. This is illustrated using the following example. TT[0] has a length of 10 MDCT Positions. This exactly matches the length of two subsequent SCBs (like 4 + 6). Then, every possible ST that is correlated with TT[0] also has a length of 10 positions. A second target portion TT[1], being adjacent to TT[0], has a length of 15 positions l (SCB having a length of 7+8). So the ST for this has a length of 15 positions instead of 10 positions as for TT[0].

[0179] If it turns out that a TT to ST cannot be located with the target portion length (when, for example, the TT length is greater than the available source range), then a correlation is not calculated and the source track is copied several times in this TT (the copy is made one after the other so that a frequency line for the lower frequency of the second copy immediately follows - in frequency - the frequency line for the higher frequency of the first copy), until the TT of the target portion is completely filled.

[0180] Next, reference is made to figure 5c which illustrates an additional preferred application of the frequency regenerator 116 of figure 1b or the IGF block 202 of figure 2a. Block 522 is a frequency portion generator that not only receives a target band ID, but additionally receives a source band ID. Exemplarily, it was determined on the encoder side that the scale factor 3 band in Figure 3a is very well suited for reconstructing the scale factor 7 band. Thus, the source bank ID would be 2 and the ID of the target band would be 7. Based on this information, the frequency portion generator 522 applies a copy or harmonic portion filling operation or any other portion filling operation to generate the raw second part of the spectral components 523. The second part of the spectral components has a frequency resolution identical to the frequency resolution included in the first set of primary spectral parts.

[0181] Then, the first spectral part of the reconstruction band as 307 of Figure 3a is input into a structure creator 524 and the second raw part 523 is also input into the structure creator 524. Then, the reconstructed structure is adjusted by the regulator 526 using a gain factor for the reconstruction band calculated by the gain factor calculator 528. Importantly, the first spectral part in the structure is not influenced by the regulator 526, but only the second raw part for the reconstruction structure is influenced. by the regulator 526. For this purpose, the gain factor calculator 528 analyzes the source band or the second raw part 523 and additionally analyzes the first spectral part in the reconstruction band to finally find the correct gain factor 527 so that the The energy of the adjusted structure emitted by the regulator 526 has the energy E4 when a band of scale factor 7 is contemplated.

[0182] In this context, it is very important to evaluate the accuracy of the high-frequency reconstruction of the present invention in comparison with HE-AAC. This is explained with respect to the scale factor 7 band in figure 3a. It is assumed that a prior art encoder as illustrated in Figure 13a would detect the spectral part 307 to be encoded with a high resolution as a “missing harmonic”. Then, the energy of this spectral component would be transmitted along with a spectral envelope information for the reconstruction band as the scale factor 7 band to the decoder. Then the decoder would recreate the missing harmonic. However, the spectral value at which the missing harmonic 307 would be reconstructed by the prior art decoder of Figure 13b would be in the middle of band 7 at a frequency indicated by the reconstruction frequency 390. Thus, the present invention avoids a frequency error 391 that would be introduced by the prior art decoder of Figure 13d.

[0183] In one implementation, the spectral analyzer is also implemented to calculate similarities between the primary spectral parts and the secondary spectral parts and to determine, based on the calculated similarities, for a second spectral part in a reconstruction range a first part spectral part that corresponds with the second spectral part as much as possible. Then, in this variable source range/destination range implementation, the parametric encoder will additionally introduce into the second encoded representation a corresponding information, indicating for each destination range a corresponding source range. On the decoder side, this information would then be used by a frequency portion generator 522 of Figure 5c that illustrates a generation of a second raw portion 523 based on a source band ID and a target band ID.

[0184] Furthermore, as illustrated in Figure 3a, the spectral analyzer is configured to analyze the spectral representation up to a maximum analysis frequency being only a small amount below half the sampling frequency and preferably being at least a quarter of the sampling frequency. sampling or typically higher.

[0185] As illustrated, the encoder operates without reducing the sampling rate and the decoder operates without increasing the sampling rate. In other words, the spectral domain audio encoder is configured to generate a spectral representation having a Nyquist frequency defined by the sampling rate of the originally input audio signal.

[0186] Furthermore, as illustrated in Figure 3a, the spectral analyzer is configured to analyze the spectral representation that starts with an initial gap-filling frequency and that ends with a maximum frequency represented by a maximum frequency included in the spectral representation, in that a spectral part extending from a minimum frequency to the initial gap-filling frequency belongs to the first set of spectral parts and wherein an additional spectral part such as 304, 305, 306, 307 having frequency values above the filling frequency gap is additionally included in the first set of primary spectral parts.

[0187] As described, the audio spectral domain decoder 112 is configured such that a maximum frequency represented by a spectral value in the first decoded representation is equal to a maximum frequency included in the time representation having the sampling rate at which the spectral value for the maximum frequency in the first set of primary spectral parts is zero or non-zero. In any case, for this maximum frequency in the first set of spectral components a scale factor for the scale factor band exists, which is generated and transmitted regardless of whether all spectral values in this scale factor band are set to zero or not. as discussed in the context of figures 3a and 3b.

[0188] The invention is therefore advantageous since, with respect to other parametric techniques for increasing compression efficiency, e.g., noise replacement and noise filling (these techniques are exclusively for efficient representation of local signal content noise type), the invention enables accurate frequency reproduction of tonal components. To date, in the state of the art, the technique directs efficient parametric representation of arbitrary signal content by spectral gap filling without the restriction of a prior division fixed into low band (LF | low band) and high band (HF | high band).

[0189] Applications of the inventive system improve upon prior art approaches and thus provide high compression efficiency, no or only minor perceptual disturbance, and full audio bandwidth even for low bit rates.

[0190] The overall system consists of • full-band central coding; • intelligent gap filling (portion filling or noise filling); • sparse tonal parts in the nucleus selected by the tonal mask; • joint stereo pair encoding for full band, including portion padding; • TNS in the portion; • spectral brightening in the IGF range.

[0191] A first step towards a more efficient system is to remove the need to transform spectral data in a second transform domain different from that of the central encoder. Since most audio codecs, such as AAC for example, use MDCT as their base transform, it is useful to perform BWE in the MDCT domain as well. A second requirement for the BWE system would be the need to preserve the tonal grid whereby HF tonal components are preserved and the quality of the encoded audio is therefore superior to existing systems. To take care of both the above mentioned requirements for a BWE scheme, a new system is proposed called Intelligent Gap Filling (IGF). Figure 2b shows the block diagram of the proposed system on the encoder side and Figure 2a shows the system on the decoder side

[0192] Subsequently, additional optional features of the first full-band frequency domain encoding processor and the full-band frequency domain decoding processor incorporating the gap-fill operation, which may be implemented together or separately, are discussed and defined.

[0193] Particularly, the spectral domain decoder 112 corresponding to block 1122a is configured to output a sequence of decoded structures of spectral values, a decoded structure being the first decoded representation, wherein the structure comprises spectral values for the first set of parts spectral values and zero indications for the secondary spectral parts. The decoding apparatus further comprises a combiner 208. Spectral values are generated by a frequency regenerator for the second set of secondary spectral parts, where both the combiner and the frequency regenerator are included within block 1122b. Thus, by combining the secondary spectral parts and the primary spectral parts, a reconstructed spectral structure comprising spectral values for the first set of primary spectral parts and the second set of spectral parts is obtained and the spectrum-time converter 118 corresponding to the IMDCT 1124 in Figure 14b then converts the reconstructed spectral structure into time representation.

[0194] As described, the spectrum-time converter 118 or 1124 is configured to perform an inverse modified discrete cosine transform 512, 514 and further comprises an overlap addition stage 516 for subsequent superposition and addition to domain structures. of time.

[0195] Particularly, the audio spectral domain decoder 1122a is configured to generate the first decoded representation such that the first decoded representation has a Nyquist frequency that defines a sampling rate being equal to a sampling rate of the generated time representation. by the spectrum-time converter 1124.

[0196] Furthermore, the decoder 1112 or 1122a is configured to generate the first decoded representation such that a first spectral part 306 is placed with respect to frequency between two secondary spectral parts 307a, 307b.

[0197] In a further application, a maximum frequency represented by a spectral value for the maximum frequency in the first decoded representation is equal to a maximum frequency included in the time representation generated by the spectrum-time converter, wherein the spectral value for the Maximum frequency in the first representation is zero or non-zero.

[0198] Furthermore, as illustrated in Figure 3, the first coded part of the audio signal further comprises a coded representation of a third set of tertiary spectral parts to be reconstructed by noise filling and the first decoding processor 1120 includes , additionally, a noise filler included in block 1122b for extracting noise filling information 308 from a coded representation of the third set of tertiary spectral parts and for applying a noise filling operation to the third set of tertiary spectral parts without using a first spectral part in a different frequency range.

[0199] Additionally, the audio spectral domain decoder 112 is configured to generate the first decoded representation having the primary spectral parts with the frequency values being greater than the frequency being equal to a frequency in the middle of the covered frequency range. by the time representation emitted by the spectrum-time converter 118 or 1124.

[0200] Additionally, the spectral analyzer or full-band analyzer 604 is configured to analyze the representation generated by the time-frequency converter 602 to determine a first set of primary spectral parts to be encoded with the first high spectral resolution and the second set of different secondary spectral parts to be encoded with a second spectral resolution that is lower than the first spectral resolution and, by means of the spectral analyzer, a first spectral part 306 is determined, with respect to frequency, between the two spectral parts secondary in figure 3 at 307a and 307b.

[0201] Particularly, the spectral analyzer is configured to analyze the spectral representation up to a maximum analysis frequency being at least one-quarter of a sampling frequency of the audio signal.

[0202] Particularly, the audio spectral domain encoder is configured to process a sequence of spectral value structures for a quantization and entropy coding, wherein, in one structure, the spectral values of the second set of secondary parts are defined at zero, or wherein, in the structure, the spectral values of the first set of primary spectral parts and the second set of secondary spectral parts are present and wherein, during subsequent processing, the spectral values in the second set of spectral parts are set to zero, as exemplarily illustrated and, 410, 418, 422.

[0203] The audio spectral domain encoder is configured to generate a spectral representation having a Nyquist frequency defined by the sampling rate of the input audio signal or the first part of the audio signal processed by the first encoding processor operating in the frequency domain.

[0204] The audio spectral domain encoder 606 is further configured to provide the first coded representation such that, for a structure of a sampled audio signal, the coded representation comprises the first set of primary spectral parts and the second set of secondary spectral parts, wherein the spectral values in the second set of spectral parts are encoded as zero or noise values.

[0205] The full-band analyzer 604 or 102 is configured to analyze the spectral representation that begins with the initial gap-filling frequency 209 and that ends with a maximum frequency fmax represented by a maximum frequency included in the spectral representation, and a spectral part extending from a minimum frequency to the initial gap-filling frequency 309 belongs to the first set of primary spectral parts.

[0206] Particularly, the analyzer is configured to apply a tonal mask processing of at least a part of the spectral representation, such that the tonal components and the non-tonal components are separated from each other, characterized by the first set of spectral parts primaries comprise the tonal components and the second set of secondary spectral parts comprises the non-tonal components.

[0207] Although the present invention has been described in the context of block diagrams, where the blocks represent real or logical hardware components, the present invention can also be implemented by a computer-implemented method. In the latter case, the blocks represent corresponding method steps where these steps represent the functionalities performed by the corresponding logical or physical hardware blocks.

[0208] Although some aspects have been described in the context of an apparatus, it is clear that these aspects also represent a description of the corresponding method, where a block or device corresponds to a method step or characteristic of a method step. Analogously, aspects described in the context of a method step also represent a description of a corresponding block or item or characteristic of a corresponding apparatus. Some or all of the steps of the method may be performed by (or using) a hardware apparatus, such as, for example, a microprocessor, a programmable computer or an electronic circuit. In some applications, one or more of the most important steps of the method may be performed by said apparatus.

[0209] The inventive encoded or transmitted signal may be stored on a digital storage medium or may be transmitted via a transmission medium, such as a cold-free transmission medium or a wired transmission medium, such as the internet.

[0210] Depending on certain implementation requirements, applications of the invention can be implemented in hardware or in software. The implementation can be carried out using a digital storage medium, for example, a floppy disk, a DVD, a Blu-Ray, a CD, a ROM memory, a PROM, an EPROM, an EEPROM or flash memory, having control signals electronically readable files stored therein, which cooperate (or may cooperate) with a programmable computer system, so that the respective method is carried out. Thus, the digital storage medium can be computer readable.

[0211] Some applications according to the invention comprise a data carrier, having electronically readable control signals, which can cooperate with a programmable computer system, so that one of the methods described herein is carried out.

[0212] Generally, applications of the present invention may be implemented as a computer program product with a program code, the program code being operative to carry out one of the methods when the computer program product is executed on a computer. Program code can, for example, be stored on a machine-readable carrier.

[0213] Other applications comprise the computer program for carrying out one of the methods described herein, stored on a machine-readable carrier.

[0214] In other words, an application of the inventive method is therefore a computer program having a program code to carry out one of the methods described here, when the computer program is executed on a computer.

[0215] Another application of the inventive method is therefore a data carrier (or a non-transitory storage medium, such as a digital storage medium or a computer-readable medium) comprising, recorded thereon, the computer program for carrying out one of the methods described here. The data carrier, digital storage medium, or recorded medium is typically tangible and/or non-transitory.

[0216] Another application of the inventive method is therefore a data stream or a sequence of signals representing the computer program for carrying out one of the methods described here. The data stream or sequence of signals may, for example, be configured to be transferred via a data communication connection, for example via the Internet.

[0217] Another application comprises a processing means, for example, a computer or a programmable logic device, configured for or adapted to perform one of the methods described herein.

[0218] Another application comprises a computer, having installed on it the computer program to carry out one of the methods described here.

[0219] Another application, according to the invention, comprises an apparatus or a system configured to transfer (e.g., electronically or optically) a computer program for carrying out one of the methods described here to a receiver. The receiver may, for example, be a computer, a mobile device, a memory device or the like. The apparatus or system may, for example, comprise a file server for transferring the computer program to the receiver.

[0220] In some applications, a programmable logic device (e.g., a field programmable gate array) may be used to perform some or all of the functionality of the methods described herein. In some applications, an array of field programmable gates may cooperate with a microprocessor in order to perform one of the methods described here. Generally, the methods are preferably performed by any hardware apparatus.

[0221] The applications described above are merely illustrative of the principles of the present invention. It is understood that modifications and variations of the arrangements and details described will be apparent to others skilled in the art. It is intended, therefore, to be limited only by the scope of the impending patent claims and not by the specific details presented in the form of a description and explanation of the applications herein.

Claims (13)

1. An audio encoder for encoding an audio signal, characterized in that it comprises: a first encoding processor (600) for encoding a first part of the audio signal in a frequency domain, wherein the first encoding processor (600) 600) comprises: a time-frequency converter (602) for converting the first part of the audio signal into a frequency domain representation, having spectral lines up to a maximum frequency of the first part of the audio signal; an analyzer (604) for analyzing the frequency domain representation up to the maximum frequency to determine primary spectral parts to be encoded with a first spectral resolution and secondary spectral parts to be encoded with a second spectral resolution, the second spectral resolution being higher lower than the first spectral resolution, wherein the analyzer (604) is configured to determine a first spectral part (306) from the primary spectral parts, the first spectral part being placed, with respect to frequency, between the two spectral parts secondary (307a, 307b) from the secondary spectral parts; a spectral encoder (606) for encoding the primary spectral parts with the first spectral resolution and for encoding the secondary spectral parts with the second spectral resolution, wherein the spectral encoder comprises a parametric encoder for calculating spectral envelope information having the second resolution spectral from the secondary spectral parts; a second encoding processor (610) for encoding a different second portion of the time domain audio signal, wherein the second encoding processor (610) comprises: a sample rate converter (900) for converting the second portion of the audio signal into a lower sample rate representation, the lower sample rate being lower than a sample rate of the audio signal, wherein the lower sample rate representation does not include the high band of the input signal ; a time domain low bandwidth encoder (910) for the time domain that encodes the lower sample rate representation; and a time domain bandwidth extension encoder (920) for parametrically encoding the high bandwidth; a controller (620) configured for analyzing the audio signal and determining which part of the audio signal is the first part of the audio signal encoded in the frequency domain and which part of the audio signal is the second part of the audio signal encoded in the time domain; and a coded signal modulator (630) for modulating a coded audio signal, comprising a first part of the coded signal for the first part of the audio signal and a second part of the coded signal for the second part of the audio signal. 2. The audio encoder according to claim 1, characterized in that it further comprises: a preprocessor (1000) configured for preprocessing the first part of the audio signal and the second part of the audio signal, wherein the preprocessor comprises: a prediction analyzer (1002) for determining prediction coefficients; and wherein the second coding processor comprises: a prediction coefficient quantizer (1010) for generating a quantized version of the prediction coefficients; and an entropy encoder for generating a coded version of the quantized prediction coefficients wherein the coded signal modulator (630) is configured to introduce the coded version into the coded audio signal. 3. The audio encoder according to any one of claims 1 or 2, characterized in that a preprocessor (1000) comprises a resampler (1004) for resampling the audio signal at a sampling rate of the second encoding processor; and wherein a prediction analyzer is configured to determine the prediction coefficients using a resampled audio signal, or wherein the preprocessor (1000) further comprises a long-term prediction analysis stage (1006) to determining one or more long-term prediction parameter(s) for the first part of the audio signal. 4. The audio coder according to any one of the preceding claims 1 to 3, characterized in that it further comprises a crossover processor (700) for calculating, from the encoded spectral representation of the first part of the audio signal, the data initialization of the second encoding processor (610), such that the second encoding processing (610) is initialized to encode the second part of the audio signal immediately after the first part of the audio signal in time in the audio signal. 5. The audio coder according to claim 4, characterized in that the crossover processor (700) comprises: a spectral decoder (701) for calculating a decoded version of the first part of the coded signal; a delay stage (707) for inputting a delayed version of the decoded version into a de-emphasis stage (617) of the second encoding processor for initialization; a weighted prediction coefficient analysis filtering block (708) for inserting an output of the filter into a codebook determiner (613) of the second encoding processor (610) for initialization; an analysis filtering stage (706) for filtering the decoded version or a pre-emphasized version (709) and for inserting a residual filter into an adaptable codebook determiner (612) of the second encoding processor for initialization; or a pre-emphasis filter (709) for filtering the decoded version and for inserting a delayed or pre-emphasized version into a synthesis filtering stage (616) of the second encoding processor (610) for initialization. 6. The audio encoder according to any one of claims 1 to 5 above, characterized in that the second coding processor comprises at least one block of the following group of blocks: a predictive analysis filter (611); an adaptive codebook stage (612); an innovative codebook stage (614); an estimator (613) for estimating an innovative codebook entry; a gain encoding/ACELP stage (615); a prediction synthesis filtering stage (616); a stage of de-emphasis (617); and a low post-filter analysis stage (618). 7. An audio decoder for decoding a coded audio signal, characterized in that it comprises: a first decoding processor (1120) for decoding a first part of the audio signal encoded in a frequency domain, the first decoding processor (1120) 1120) comprising: a spectral decoder (1122) for decoding the primary spectral parts with a high spectral resolution and for synthesizing the secondary spectral parts using a parametric representation of the secondary spectral parts and at least one first decoded spectral part to obtain a representation decoded spectral, wherein the spectral decoder (1122) is configured to generate the first decoded representation such that a first spectral part (306) is placed with respect to the frequency between the two secondary spectral parts (307a, 307b); and a time-frequency converter (1120) for converting the decoded spectral representation into a time domain to obtain a first part of the decoded audio signal; a second decoding processor (1140) for decoding a second part of the time domain encoded audio signal to obtain a second part of the decoded audio signal; and a combiner (1160) for combining the first decoded spectral part and the second decoded spectral part to obtain a decoded audio signal; wherein the second decoding processor comprises: a low-band time-domain decoder (1200) for decoding a low-band time-domain signal; a sample rate amplifier (1210) for increasing the sample rate of the low band time domain signal; a time domain bandwidth extension decoder (1220) for synthesizing a high bandwidth of a time domain output signal; and a mixer (1230) for mixing a synthesized high-band time domain signal and an expanded low-band time domain signal. 8. The audio decoder of claim 7, wherein the sample rate amplifier (1210) comprises an analysis filter bank (1471) operating at a first sample rate of the low-band time domain decoder and a synthesis filter bank (1473) that operates at a second output sample rate being higher than the first time domain low-band sample rate. 9. The audio decoder according to any one of claims 7 or 8, characterized in that the first decoding processor (1120) comprises a long-term predictive adaptive post-filter (1420) for post-filtering the first part of the decoded signal , wherein the filter (1420) is controlled by one or more long-term prediction parameter(s) included in the encoded audio signal. 10. The audio decoder according to any one of claims 7 to 9, characterized in that the crossover processor (1170) comprises: a delay stage (1172) for delaying the first part of the additional decoded signal and for inserting a delayed version of the first part of the decoded signal in a de-emphasis stage (1144) of the second decoding processor for initialization; a pre-emphasis filter (1173) and a delay stage (1175) for filtering and delaying the first part of the additional decoded signal and for inserting an output delay stage into a preview synthesis filter (1143) of the second processing processor. decoding for boot; a prediction analysis filter (1174) for generating a prediction residual signal of the first additional decoded spectral part or an additional first pre-emphasized decoded signal part (1173) and for inserting a residual prediction signal into a signal synthesizer. codebook (1141) of the second decoding processor (1200); or a switch (1480) for inserting the first part of the additional decoded signal into an analysis stage (1471) of a resampler (1210) of the second decoding processor for initialization. 11. The audio decoder according to any one of claims 7 to 10, characterized in that the second decoding processor (1200) comprises at least one block of the group of blocks comprising: an ACELP for decoding gains and a book of innovative codes; an adaptive codebook synthesis stage (1141); an ACELP postprocessor (1142); a prediction synthesis filter (1143); and a de-emphasis stage (1144). 12. A method for encoding an audio signal, characterized by comprising: a first encoding (600) of a first part of the audio signal in a frequency domain, wherein the first encoding (600) comprises: conversion (602) of the first part of the audio signal in a frequency domain representation, having spectral lines up to a maximum frequency of the first part of the audio signal; analysis (604) of the frequency domain representation up to the maximum frequency to determine the primary spectral parts to be encoded with a first spectral resolution and secondary spectral parts to be encoded with a second spectral resolution, the second spectral resolution being lower than the first spectral resolution, wherein the analysis (604) determines a first spectral part (306) from the primary spectral parts, the first spectral part being placed, with respect to frequency, between the two secondary spectral parts (307a, 307b) from the secondary spectral parts; coding (606) of the primary spectral parts with the first spectral resolution and for coding the secondary spectral parts with the second spectral resolution, wherein coding the second spectral part comprises calculating, from the secondary spectral parts, the spectral envelope information having the second spectral resolution; second encoding (610) of a second part of the audio signal different in the time domain, which the second encoding (610) comprises; converting (900) the second part of the audio signal to a lower sampling rate representation, the lower sampling rate being lower than the sampling rate of the audio signal, wherein the higher sampling rate representation low does not include the high band of the input signal; time domain encoding (910) of the lowest sampling rate representation; and high band parametric coding (920); analyzing (620) the audio signal and determining which part of the audio signal is the first part of the audio signal encoded in the frequency domain and which part of the audio signal is the second part of the audio signal encoded in the time domain ; and modulating (630) a coded audio signal, comprising a first part of the coded signal for the first part of the audio signal and a second part of the coded signal for the second part of the audio signal. 13. Method of decoding a coded audio signal, characterized by comprising: a first decoding (1120) of a first part of the coded audio signal in a frequency domain, the first decoding (1120) comprising: decoding (1122) of primary spectral parts with a high spectral resolution and synthesizing the secondary spectral parts using a parametric representation of the secondary spectral parts and at least one first decoded spectral part to obtain a decoded spectral representation, wherein the decoding (1122) comprises generating of the first decoded representation such that a first spectral part (306) is placed with respect to the frequency between the two secondary spectral parts (307a, 307b); and converting (1120) the decoded spectral representation into a time domain to obtain a first part of the decoded audio signal; second decoding (1140) of a second part of the time domain encoded audio signal to obtain a second part of the decoded audio signal, and combining (1160) the first decoded spectral part and the second decoded spectral part to obtain a decoded audio, wherein the second decoding comprises: decoding the low band time domain to obtain a low band time domain signal; low-band time domain signal broadening; high-band synthesis of the time domain output signal; and mixing a synthesized high band time domain signal and an extended low band time domain signal.