RU2804032C1 - Audio signal processing device for stereo signal encoding into bitstream signal and method for bitstream signal decoding into stereo signal implemented by using audio signal processing device - Google Patents

Abstract

FIELD: computer engineering.

SUBSTANCE: processing audio data. The technical result is achieved by analysing the stereo input signal by applying both the coding of the middle/side stereo signal and the coding of the left/right stereo signal, and selecting either the coding mode of the middle/side stereo signal or the coding mode of the left/right stereo signal based on the estimated entropy for each stereo encoding mode, encoding the stereo input signal using the selected stereo encoding mode in the first frequency band to obtain an encoded stereo signal in the first frequency band, downmixing the stereo input signal to mono in the second frequency band, encoding the mono signal in a second frequency band using transform coding to obtain an encoded mono signal in the second frequency band and output the encoded stereo signal in the first frequency band and the encoded mono signal in the second frequency band as an encoded output signal.

EFFECT: providing the possibility of combining PS-coding using the residual signal and adaptive perceptual stereo L/R-or M/S-coding.

23 cl, 5 dwg

Description

Field of technical application

The application relates to the coding of audio signals, in particular to the coding of stereo audio signals, combining parametric coding and waveform coding technologies.

BACKGROUND OF THE INVENTION

Joint encoding of the left (L) and right (R) channels of a stereo signal allows for more efficient encoding compared to independent encoding of L and R. The traditional approach to joint stereo encoding is mid/side (M/S) encoding. In this case, the average (M) signal is formed by adding the L and R signals, for example, the M signal can have the form

.

Similarly, the side (S) signal is formed by subtracting the L and R channels, for example, the S signal has the form

.

In the case of M/S encoding, M and S signals are encoded instead of L and R signals.

In the MPEG (Moving Picture Experts Group) AAC (Advanced Audio Coding) standard (see ISO/IEC standard document 13818-7), stereo L/R coding and stereo M/S coding can be selected as time-dependent and frequency-dependent way. Thus, a stereo encoder may use L/R coding for some frequency bands of a stereo signal, while M/S coding is used to encode other frequency bands of a stereo signal (frequency dependent method). In addition, the encoder can switch between L/R encoding and M/S encoding over time (time-dependent manner). In MPEG AAC, stereo coding is performed in the frequency domain, more specifically the MDCT (Modified Discrete Cosine Transform) domain. This allows adaptive selection between L/R coding and M/S coding in a frequency-dependent as well as time-dependent manner. The decision to choose between stereo L/R encoding and stereo M/S encoding can be based on an evaluation of the side signal: when the energy of the side signal is low, stereo M/S encoding should be used, which is more efficient. In an alternative to deciding between two encoding schemes, both schemes may be tested and the choice may be based on the resulting quantization attempts, i.e. on observed perceptual entropy.

An alternative approach to joint stereo coding is parametric stereo (PS) coding. In this case, the stereo signal is transmitted as a monaural downmix signal after encoding the downmix signal with a conventional audio encoder such as an AAC encoder. The downmix signal is a superposition of the L and R channels. The monaural downmix signal is transmitted in combination with additional time- and frequency-dependent PS encoding parameters such as inter-channel intensity difference (IID) and inter-channel cross-correlation (ICC). In the decoder, based on the decoded downmix signal and parametric stereo encoding parameters, a stereo signal is reconstructed that is an approximation of the perceived spatial image of the original stereo signal. For reconstruction, the decorrelator generates a decorrelated version of the downmixed signal. Said decorrelator can be implemented by means of a corresponding phase filter. PS encoding and decoding is described in the article "Low Complexity Parametric Stereo Coding in MPEG-4", H. Purnhagen, Proc. Of the 7th Int. Conference on Digital Audio Effects (DAFx'04), Naples, Italy, October 5-8, 2004, pages 163-168. The disclosures from this document are incorporated by reference herein.

The MPEG Surround standard (see document ISO/IEC 23003-1) uses the concept of PS encoding. In an MPEG Surround decoder, many output channels are created based on fewer input channels and control parameters. MPEG Surround decoders and encoders are built by cascading parametric stereo modules, which in the MPEG Surround standard are called OTT modules (one-to-two modules) for the encoder, and R-OTT modules (reverse one-to-two modules) for coder. The OTT module defines two output channels through a single input channel (downmixed signal) accompanied by PS encoding parameters. The OTT module corresponds to the PS decoder, and the R-OTT module corresponds to the PS encoder. Parametric stereo coding can be implemented using MPEG Surround with one OTT module on the decoder side and one R-OTT module on the encoder side; this is also called "MPEG Surround 2-1-2" mode. The bitstream syntax may be different, but the underlying theory and signal processing are the same. Therefore, in the following description, all references to PS encoding also include parametric stereo encoding based on "MPEG Surround 2-1-2" or MPEG Surround.

A PS encoder (eg, an MPEG Surround PS encoder) may define a residual signal ( RES ) that is transmitted in addition to the downmix signal. This residual signal indicates the error associated with the display of the original channels through their downmixing and PS encoding parameters. In the decoder, the residual signal can be used instead of a decorrelated version of the downmixed signal. This allows a better reconstruction of the signal shapes of the original L and R channels. The use of an additional residual signal is described, for example, in the MPEG Surround standard (see document ISO/EEC 23003-1) and in the article "MPEG Surround - The ISO/MPEG Standard for Efficient and Compatible Multi-Channel Audio Coding”, J. Herre et al., Audio Engineering Convention Paper 7084, ^122nd Convention, May 5-8, 2007. Disclosures from both documents, in particular, remarks therein relating to residual signal, by reference are included in this description.

PS encoding with residual signal is a more general approach to co-encoding a stereo signal than M/S encoding: M/S encoding performs signal rotation when converting L/R signals to M/S signals. Residual PS encoding also performs signal rotation when converting L/R signals into downmix and residual signals. However, in the latter case, the rotation of the signal is variable and depends on the PS encoding parameters.

Because of the greater generality of the residual PS coding approach, PS residual coding can encode certain types of signals, such as a panned mono signal, more efficiently than M/S coding. Thus, the proposed encoder makes it possible to effectively combine parametric stereo coding technologies with stereo coding technologies based on waveform coding.

Often, perceptual stereo encoders, such as the MPEG AAC perceptual stereo encoder, may decide between L/R stereo encoding and M/S stereo encoding, where in the latter case a mid/side signal is generated based on the stereo signal. This choice may depend on the frequency, i.e. Some frequency bands may use stereo L/R encoding, while other frequency bands may use stereo M/S encoding.

In a situation where the L and R channels are essentially independent signals, said perceptual stereo encoder will generally not use stereo M/S coding because in this situation this coding scheme does not provide any coding improvement over stereo L /R-coding. The encoder will fall back to the level of simple stereo L/R encoding, essentially processing L and R independently.

In the same situation, the PS encoder system will produce a downmix signal that contains both the L and R channels, preventing the L and R channels from being processed independently. For PS encoding with a residual signal, this may mean less efficient encoding compared to stereo encoding. where stereo L/R encoding or stereo M/S encoding is adaptively selected.

Thus, there are situations in which a PS encoder is superior to a perceptual stereo encoder with an adaptive choice between stereo L/R encoding and stereo M/S encoding, while in other situations the latter encoder is superior to a PS encoder.

Brief description of the invention

The present application describes an audio encoder system and encoding method that is based on the idea of combining PS encoding using a residual signal and adaptive perceptual stereo L/R or M/S encoding (eg, perceptual joint AAC stereo encoding in the MDCT domain). This allows you to combine the benefits of adaptive stereo L/R or M/S encoding (eg used in MPEG AAC) with the benefits of residual PS coding (eg used in MPEG Surround). In addition, the application describes a corresponding audio decoder system and a decoding method.

The first aspect of the application relates to an encoder system for encoding a stereo signal into a bitstream signal. According to one embodiment of the encoder system, the encoder system includes a downmixing step for generating a downmix signal and a residual signal from the stereo signal. The residual signal may cover all or only part of the usable frequency range of the audio signal. In addition, the encoder system includes a parameter determination step for determining PS encoding parameters such as intensity difference between channels and cross-correlation between channels. Preferably, the PS encoding parameters are frequency dependent. These downmixing and parameter determination stages typically form part of the PS encoder.

In addition, the encoder system includes perceptual encoding means downstream of the downmixing stage, where a choice is made between two encoding schemes:

- coding based on the sum of the downmix signal and the residual signal and based on the difference between the downmix signal and the residual signal; or

- coding based on the downmixed signal and based on the residual signal.

It should be noted that in the case of coding based on the downmix signal and the residual signal, the downmix signal and the residual signal may be encoded, or signals proportional to these signals may be encoded. In the case of sum-difference encoding, the sum and difference may be encoded, or signals proportional to those signals may be encoded.

The selection may be frequency dependent (or time dependent), i.e. for the first frequency band, coding based on the sum signal and the residual signal may be chosen, while for the second frequency band, coding based on the downmix signal and the residual signal may be chosen.

Such an encoder system has the advantage of being able to switch between stereo L/R encoding and PS encoding with a residual signal (preferably in a frequency dependent manner): if the perceptual encoding means makes a choice (for a specific band or for the entire frequency range) in favor of encoding based on downmix and residual signals, the encoding system behaves like a system using standard PS encoding with a residual signal. However, if the perceptual encoding means makes a choice (for a particular band or for an entire frequency range) to encode based on the sum of the downmix signal and the residual signal and the difference signal of the downmix signal and the residual signal, under certain conditions, the addition and subtraction operations are largely compensate for the previous downmix operation (except in the case of a possibly different gain), and thus the overall system can effectively perform L/R encoding on the entire stereo signal or on one of its frequency bands. For example, such conditions occur when the L and R channels of a stereo signal are independent and at the same level, which will be explained in more detail later.

Preferably, the adaptation of the coding scheme is time dependent and frequency dependent. Thus, preferably, some frequency bands of the stereo signal are encoded by an L/R coding scheme, while other frequency bands of the stereo signal are encoded by a PS coding scheme with a residual signal.

It should be noted that in the case of coding based on the downmix signal and the residual signal, as discussed above, the actual signal that is input to the base encoder can be generated by two successive operations on the downmix signal and the residual signal, which are inverse (except in the case , possibly with a different gain). For example, the downmix signal and the residual signal are supplied to the M/S to L/R conversion stage, and then the output signal of the conversion stage is supplied to the L/R to M/S conversion stage. The resulting signal (which is then used for encoding) corresponds to the downmix signal and the residual signal (except in the case of a possibly different gain).

The following embodiment of the invention uses this idea. According to one embodiment of an encoder system, the encoder system, as discussed above, includes a downmixing step and a parameter determination step. In addition, the encoder system includes a transformation step (eg, part of the encoding means described above). The conversion step generates a pseudo-L/R stereo signal by performing conversion on the downmix signal and the residual signal. The transform step preferably performs sum-and-difference conversions, where the downmix signal and the residual signal are added, generating one channel of pseudo-stereo signal (the sum may be multiplied by some factor), and subtracted from each other, generating a second channel of pseudo-stereo signal ( the difference may be multiplied by some coefficient). Preferably, the first channel (eg, pseudo-left channel) of the pseudo-stereo signal is proportional to the sum of the downmix and residual signals, where the second channel (eg, pseudo-right channel) of the pseudo-stereo signal is proportional to the difference of the downmix and residual signals. Thus, the _downmixed DMX signal and the residual RES signal from the PS encoder can be converted into a pseudo-stereo signal Lp , Rp according to the following _equations :

L _p = g ( DMX + RES )

R _p = g ( DMX – RES ).

In the above equations, the normalization gain g , for example, has the value .

The pseudo-stereo signal is preferably processed by a perceptual stereo encoder (eg, forming part of the encoding means). For encoding, select either stereo L/R encoding or stereo M/S encoding. The adaptive perceptual stereo L/R or M/S encoder may be an AAC-based encoder. Preferably, the choice between stereo L/R encoding and stereo M/S encoding depends on frequency; thus, the choice may vary for different frequency bands, as discussed above. Moreover, the choice between L/R encoding and M/S encoding is preferably time dependent. The decision to choose between L/R encoding and M/S encoding is preferably made by the perceptual stereo encoder.

Said perceptual stereo encoder for M/S encoding has the capability of internally computing (pseudo-) M and S signals (in the time domain or in selected frequency bands) based on the pseudo-stereo L/R signal. The indicated M and S signals correspond to the downmix and residual signals (except in the case of a possibly different gain). Thus, if a perceptual stereo encoder opts for M/S encoding, it is effectively encoding the downmix and residual signals (which correspond to the pseudo M and S signals) as it would be done in a system using standard PS encoding with residual signal.

Moreover, under specific conditions, the conversion step largely compensates for the previous downmix operation (except in the case of possibly different gain), and thus the encoder system as a whole can effectively perform L/R encoding of the entire stereo signal or one from its frequency bands (if L/R coding is selected as the perceptual encoder). This case corresponds, for example, to the case where the L and R channels of a stereo signal are independent and have the same level, which will be explained in more detail later. Thus, for a given frequency band, a pseudo-stereo signal is substantially equal to or proportional to the stereo signal if, for that frequency band, the left and right channels of the stereo signal are substantially independent and have substantially the same level.

Thus, the encoder system effectively allows switching between stereo L/R encoding and PS encoding with a residual signal in order to be able to adapt to the properties of a given stereo input signal. Preferably, the adaptation of the coding scheme is time dependent and frequency dependent. Thus, preferably, some frequency bands of the stereo signal are encoded by an L/R coding scheme, while other frequency bands of the stereo signal are encoded by a PS coding scheme with a residual signal. It should be noted that M/S encoding is essentially a special case of PS encoding with a residual signal (since the L/R to M/S conversion is a special case of the downmixing operation of PS encoding), and thus the system encoder can also perform full M/S encoding.

This embodiment of the invention comprising a transform step downstream of a PS encoder and upstream of a perceptual stereo L/R or M/S encoder has the advantage that a conventional PS encoder and traditional perceptual encoder. However, the PS encoder, or perceptual encoder, in this case can be adapted for specific use.

The new concept improves the efficiency of stereo encoding, allowing PS encoding and co-stereo encoding to be effectively combined.

According to an alternative embodiment of the invention, the encoding means as discussed above include a transform step for performing sum and difference conversion based on the downmix signal and the residual signal for one or more frequency bands (e.g., for the entire usable frequency range or just for one frequency band). This conversion can be performed in the frequency domain or in the time domain. The conversion stage generates a pseudo-left/right stereo signal for one or more frequency bands. One channel of the pseudo-stereo signal corresponds to the sum, and the second channel of the pseudo-stereo signal corresponds to the difference.

Thus, in the case where encoding is based on the sum and difference signals, the output signal of the transform step can be used for encoding, while in the case where encoding is based on the downmix signal and the residual signal, signals in the upstream direction relative to coding stage. Thus, this embodiment of the invention does not use two sequential sum and difference transformations on the downmix signal and the residual signal, resulting in a downmix signal and a residual signal (except in the case of a possibly different gain).

When selecting encoding based on the downmix signal and the residual signal, parametric stereo encoding of the stereo signal is selected. When selecting sum-difference encoding (ie, pseudo-stereo encoding), L/R encoding of the stereo signal is selected.

The transform stage may be an L/R to M/S transform stage forming part of a perceptual encoder with adaptive selection between stereo L/R and M/S encoding (perhaps as opposed to a traditional L/R to M/S transform stage , the gains are different). It should be noted that the decision to choose between stereo L/R and M/S encoding must be reversible. Thus, encoding based on the downmix signal and the residual signal is selected (i.e., the encoded signal does not go through the transform step) when the decision makers decide on perceptual M/S decoding, and encoding based on pseudo-stereophonic signal generated by the transform step is selected (i.e., the encoded one goes through the transform step) when the decision makers decide on perceptual L/R decoding.

The encoder system according to any of the above-discussed embodiments of the invention may include an additional SBR (spectral band replication) encoder. SBR is a form of HFR (high frequency reconstruction). The SBR encoder determines additional information to reconstruct the higher frequency range of the audio signal at the decoder. Only the lower frequency range is encoded by the perceptual encoder, which reduces the data bit rate. Preferably, the SBR encoder is connected upstream of the PS encoder. Thus, the SBR encoder can be in the stereo domain and generate SBR parameters for the stereo signal. This will be discussed in more detail in connection with the graphic materials.

Preferably, the PS encoder (ie, the downmixing step and the parameter determination step) performs operations in the upsampled frequency domain (the PS encoder, which is discussed below, also preferably performs operations in the oversampled frequency domain). For time domain to frequency domain conversion upstream of the PS encoder, a hybrid complex valued filter bank containing a QMF (quadrature mirror filter) and a Nyquist filter can be used, as described in the MPEG Surround standard (see document ISO/IEC 23003-1). This allows adaptive signal processing in time and frequency without audible aliasing artifacts. Adaptive L/R or M/S encoding, on the other hand, is preferably performed in the critically sampled MDCT domain (e.g., as described in AAC) in order to provide an efficient representation of the quantized signal.

The conversion between the downmix and residual signals and the pseudo-L/R stereo signal can be done in the time domain, since the PS encoder and the perceptual stereo encoder are typically coupled in the time domain anyway. Thus, the conversion stage for generating the pseudo-L/R signal can perform operations in the time domain.

In other embodiments of the invention, which are discussed in connection with the drawings, the transform stage performs operations in the resampled frequency domain or the critically sampled MDCT domain.

The second aspect of the application relates to a decoder system for decoding a bitstream signal generated by the encoder system discussed above.

According to one embodiment of a decoder system, the decoder system includes perceptual decoding means for decoding based on a bitstream signal. The decoding means are configured to generate, by decoding, the (internal) first signal and the (internal) second signal, and to output the downmix signal and the residual signal. The downmix signal and the residual signal are selectively based on

- the sum of the first signal and the second signal and the difference of the first signal and the second signal

or

- on the first signal and on the second signal.

As discussed above in connection with the encoder system, in this case, the selection may also be frequency dependent and frequency independent.

In addition, the system includes an upmixing step for generating a stereo signal based on the downmix signal and the residual signal, where the upmixing operation of the upmixing step depends on one or more parametric stereo encoding parameters.

Similar to the encoder system, the decoder system actually allows switching between L/R decoding and PS decoding with a residual signal, preferably in a time-dependent and frequency-dependent manner.

According to another embodiment of the invention, the decoder system includes a perceptual stereo decoder (eg, forming part of the decoding means) for decoding the bitstream signal, where the decoder generates a pseudo-stereo signal. The perceptual decoder may be an AAC-based decoder. For a perceptual stereo decoder, perceptual L/R decoding or perceptual M/S decoding is selected in a frequency-dependent or frequency-independent manner (the actual selection is preferably controlled by the encoder's decision, which is carried as side information in the bitstream). The decoder makes the choice of a decoding scheme based on the encoding scheme used for encoding. The encoding scheme to be used may be indicated to the decoder by information contained in the received bit stream.

In addition, a conversion step is provided for generating the downmix signal and the residual signal by performing conversion of the pseudo-stereo signal. In other words, the pseudo-stereo signal received from the perceptual decoder is converted back into the downmix and residual signals. This conversion is a sum-difference conversion: the resulting downmix signal is proportional to the sum of the left channel and the right channel of the pseudo-stereo signal, the resulting residual signal is proportional to the difference of the left channel and the right channel of the pseudo-stereo signal. Thus, a quasi-conversion of L/R to M/S is carried out. A pseudo-stereophonic signal with two channels L _p , R _p can be converted into downmix and residual signals in accordance with the following equations:

.

In the above equations, the normalization gain g can, for example, have the value . The residual signal RES used in the decoder may cover the entire usable frequency range of the audio signal or only a portion of the usable frequency range of the audio signal.

The downmix and residual signals are then processed in the upmixing stage of the PS decoder to produce the final stereo output signal. The upmixing of the downmix and residual signals into a stereo signal depends on the received PS encoding parameters.

According to an alternative embodiment of the invention, the perceptual decoding means may include a sum and difference transform step for performing a transform based on the first signal and the second signal for one or more frequency bands (eg, for the entire frequency range in use). Thus, the conversion step generates a downmix signal and a residual signal in the case where the downmix signal and the residual signal are based on the sum of the first signal and the second signal and on the difference of the first signal and the second signal. The conversion stage can perform operations in the time domain or in the frequency domain.

As similarly discussed in connection with the encoder system, the transform stage may be an M/S to L/R transform stage forming part of a perceptual decoder with adaptive selection between stereo L/R and M/S decoding (possibly a gain different from the traditional M/S to L/R conversion step). It should be noted that the choice between stereo L/R and M/S decoding must be reversible.

The decoder system according to any of the preceding embodiments of the invention may include an additional SBR decoder for decoding additional information from the SBR encoder and for generating a high frequency component of the audio signal. Preferably, the SBR decoder is located downstream of the PS decoder. This will be discussed in more detail in connection with the graphic materials.

Preferably, the upmixing step performs operations in the resampled frequency domain, for example, a hybrid filter bank may be used upstream of the PS decoder as described above.

The L/R to M/S conversion can be performed in the time domain since the perceptual decoder and the PS decoder (including the upmixing stage) are typically coupled in the time domain.

In other embodiments of the invention described in connection with the drawings, the L/R to M/S conversion is performed in the oversampled frequency domain (eg, QMF) or in the critically sampled frequency domain (eg, MDCT).

The third aspect of the application relates to a method for encoding a stereo signal into a bitstream signal. The method performs operations similar to the encoder system discussed above. Thus, the above remarks relating to the encoder system generally also apply to the encoding method.

A fourth aspect of the application relates to a method for decoding a bitstream signal including PS encoding parameters to generate a stereo signal. The method performs operations in the same way as the decoder system discussed above. Thus, the above remarks regarding the decoder system generally also apply to the decoding method.

The invention will be explained below by way of illustrative examples with reference to the accompanying drawings, in which

fig. 1 illustrates one embodiment of an encoder system, where optionally PS encoding parameters assist psychoacoustic control in a perceptual stereo encoder;

fig. 2 0 one of the embodiments of the PS encoder;

fig. 3 shows one embodiment of a decoder system;

fig. 4 illustrates one further embodiment of a PS encoder including a detector for disabling PS encoding if L/R encoding is preferred;

fig. 5 illustrates one embodiment of a conventional PS encoder system including an additional SBR encoder for the downmix signal;

fig. 6 illustrates one embodiment of an encoder system comprising an additional SBR encoder for the downmix signal;

fig. 7 illustrates one embodiment of an encoder system comprising an additional SBR encoder in the stereo domain;

fig. 8a-8d are different time-frequency representations of one of the two output channels at the decoder output;

fig. 9a is one embodiment of a basic encoder;

fig. 9b illustrates one embodiment of an encoder that allows switching between linear prediction domain coding (typically for mono signals only) and transform domain coding (typically for both mono and stereo signals);

fig. 10 illustrates one embodiment of an encoder system;

fig. 11a is part of one embodiment of an encoder system;

fig. 11b is an illustrative implementation of the embodiment of FIG. 11a;

fig. 11c is an alternative to the embodiment of FIG. 11a;

fig. 12 illustrates one embodiment of an encoder system;

fig. 13 is one embodiment of a stereo encoder forming part of the encoder system of FIG. 12;

fig. 14 illustrates one embodiment of a decoder system for decoding a bitstream signal generated by the encoder system of FIG. 6;

fig. 15 illustrates one embodiment of a decoder system for decoding a bitstream signal generated by the encoder system of FIG. 7;

fig. 16a is part of one embodiment of a decoder system;

fig. 16b is an illustrative implementation of the embodiment of FIG. 16a;

fig. 16c is an alternative to the embodiment of FIG. 16a;

fig. 17 illustrates one embodiment of an encoder system; And

fig. 18 illustrates one embodiment of a decoder system.

In fig. 1 illustrates an embodiment of an encoder system that combines residual PS encoding and adaptive perceptual stereo L/R or M/S encoding. This embodiment of the invention is merely an illustration of the principles of the present application. It should be understood that modifications and changes to this embodiment of the invention will be apparent to those skilled in the art. The encoder system includes PS encoder 1 receiving the L, R stereo signal. PS encoder 1 contains a downmixing stage designed to generate the DMX downmix and residual RES signals from the L, R stereo signal. This operation can be described using the matrix H ^{– 1} H ^–1 2x2 downmix that converts the L and R signals into a DMX downmix and a RES residual signal:

.

Typically, the H ^–1 matrix is frequency dependent or time dependent, i.e. the elements of the H ^–1 matrix change with frequency and change from one time slice to another. The H ^–1 matrix may be updated with each frame (eg, every 21 or 48 ms) and may have a frequency resolution of multiple bands, such as 28, 20, or 10 bands (called "parameter bands") on a perceptually oriented frequency scale (such as the Bark scale ).

The elements of the H ^–1 matrix depend on time- and frequency-dependent PS encoding parameters: IID (intensity difference between channels; also called CLD - channel level difference) and ICC (cross-correlation between channels). To determine the PS encoding parameters 5, for example, IID and ICC, the PS encoder 1 includes a parameter determination step. An example of calculating the matrix elements of the inverse matrix H has the following form and is described in the MPEG Surround standard specification, document ISO/IEC 23003-1, subclause 6.5.3.2, which is incorporated by reference into this description:

,

Where

, And ,

Where

,

and where .

In addition, the encoder system includes a conversion step 2 that converts the downmix signal DMX and the residual signal RES from the PS encoder 1 into a pseudo-stereo signal L _p , R _p , for example, in accordance with the following equations:

.

In the above equations, the normalization gain g has, for example, the value . For the two pseudo-stereo signal equations L _p , R _p can be rewritten as:

.

The pseudo-stereo signal L _p , R _p is then supplied to the perceptual stereo encoder 3, which adaptively selects between L/R stereo encoding and M/S stereo encoding. M/S coding is a form of joint stereo coding. L/R coding may also be based on some aspects of joint coding, for example, bits for L and R channels may be jointly allocated from a common bit store.

The choice between stereo L/R or M/S encoding is preferably frequency dependent, i.e. Some frequency bands may be L/R encoded, while other frequency bands may be M/S encoded. One implementation of the choice between stereo L/R or M/S encoding is described in the document "Sum-Difference Stereo Transform Coding", J. D. Johnston et al., IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) 1992 , pages 569–572. The discussion of the choice between stereo L/R or M/S encoding in this document, particularly sections 5.1 and 5.2, is incorporated by reference herein.

Based on the pseudo-stereo signal L _p , R _{p ,} the perceptual encoder 3 can perform an internal calculation of the (pseudo-) mid/side signals M _p , Sp _. These signals essentially correspond to the downmix DMX signal and the residual RES signal (except perhaps with a different gain). Then, if the perceptual encoder 3 chooses to M/S encode a frequency band, then the perceptual encoder 3 essentially encodes for that frequency band the downmixed DMX signal and the residual RES signal (except in the case of a possibly different gain) as this can also be performed in a traditional perceptual encoder system using traditional PS encoding with a residual signal. PS encoding parameters 5 and the output bit stream 4 of the perceptual encoder 3 are multiplexed into a single bit stream 6 by means of a multiplexer 7.

In addition to PS encoding the stereo signal, the encoder system of FIG. 1, as will be described below, allows L/R encoding of a stereo signal. As discussed above, the elements of the encoder's downmix matrix H ^–1 (as well as the elements of the upmix matrix H used in the decoder) depend on time- and frequency-dependent PS encoding parameters: IID (intensity differences between channels; also called CLD - difference channel levels) and ICC (inter-channel cross-correlation). An example of calculating the matrix elements of the upmixing matrix H is described above. In the case of using residual signal encoding, the right column of the 2x2 upmix matrix H is:

.

However, instead, the right column of the 2x2 matrix H should preferably be modified as

.

The left column is preferably calculated as given in the MPEG Surround specification.

Modifying the right column of the upmix matrix H ensures that when IID = 0 dB and ICC = 0 (i.e., in the case where for the corresponding band the stereo channels L and R are independent and have the same level), the following matrix is obtained for the band Upmixing :

.

It should be noted that the upmix matrix H , as well as the downmix matrix H ^–1 , are typically frequency dependent and time dependent. Thus, the matrix values are different for different time/frequency tiles (a tile corresponds to the intersection of a specific frequency band and a specific time period). In the case discussed above, the matrix H ^–1 of the down-mix is identical to the matrix H of the up-mix. Thus, the pseudo-stereo signal L _p , R _p for a band can be calculated using the following equation:

.

Thus, in this case, residual PS encoding using the H ^–1 downmix matrix followed by generation of the L/R pseudo signal in transform step 2 corresponds to the identity matrix and does not change the stereo signal at all for the corresponding frequency band, i.e. .e.

.

That is, transform step 2 compensates the downmix matrix H ^–1 so that the pseudo-stereo signal L _p , R _p matches the input stereo signal L , R .

This allows the original L, R stereo input signal to be encoded by the perceptual encoder 3 for a specific band. When L/R encoding is selected by perceptual encoder 3 to encode a particular band, the encoder system behaves as a perceptual L/R encoder designed to encode the L, R band of the stereo input signal.

The encoder system of FIG. 1 allows smooth adaptive switching between L/R encoding and PS encoding with a residual signal in a frequency- and time-dependent manner. The encoder system avoids inhomogeneities in the waveform when switching between encoding schemes. This prevents artifacts from appearing. In order to achieve smooth transitions to the elements of the H ^–1 matrix in the encoder and the H matrix in the decoder, linear interpolation can be used for discrete values between two modifications of stereo parameters.

In fig. 2 shows one embodiment of the PS encoder 1. The PS encoder 1 includes a downmixing step 8 which, based on the stereo signal L, R, generates a downmix signal DMX and a residual signal RES . In addition, the PS encoder 1 includes a parameter estimation step 9 for estimating PS encoding parameters 5 based on the stereo signal L, R.

In fig. 3 shows one embodiment of a corresponding decoder system configured to decode a bitstream 6 generated by the encoder system of FIG. 1. This embodiment of the invention is merely an illustration of the principles of the present application. It should be understood that modifications and changes to this embodiment of the invention will be apparent to those skilled in the art. The decoder system includes a demultiplexer 10 for separating the PS encoding parameters 5 and the audio bitstream 4 generated by the perceptual encoder 3. The audio bitstream 4 is supplied to a perceptual stereo decoder 11, which is capable of selectively decoding the L/R-encoded bitstream, or M/S-coded audio bitstream. The operations of the decoder 11 are inverse to those of the encoder 3. Similar to the perceptual encoder 3, the perceptual decoder 11 preferably allows for a frequency-dependent and time-dependent encoding scheme. Some frequency bands L/R encoded by encoder 3 are L/R decoded by decoder 11, while other frequency bands M/S encoded by encoder 3 are M/S decoded by decoder 11. Decoder 11 outputs the pseudo-stereo signal L _p , R _p , which was previously the input signal of the perceptual encoder 3. The pseudo-stereo signal L _p , R _p obtained from the perceptual decoder 11 is converted back into the downmix signal DMX and the residual signal RES through step 12 L/R to M/S conversions. The operations of the L/R to M/S conversion step 12 on the decoder side are the reverse of the operations of the conversion step 2 on the encoder side. Preferably, the conversion step 12 determines the downmix signal DMX and the residual signal RES in accordance with the following equations:

.

In the above equations, the normalization gain g is identical to the normalization gain g on the encoder side and, for example, has the value .

The downmix signal DMX and the residual signal RES are then processed by the PS decoder 13 to obtain the final output signals L and R. The upmixing step of the decoding process in the case of PS encoding with a residual signal can be described using a 2x2 upmix matrix H , which converts the downmix DMX signal and residual RES signal back to L and R channels:

.

The calculation of the elements of the upmix matrix H has already been discussed above.

The PS encoding and PS decoding processes in the PS encoder 1 and PS decoder 13 are preferably carried out in the resampled frequency domain. For time-frequency conversion upstream of the PS encoder, a hybrid complex-valued filter bank containing a QMF (quadrature mirror filter) and a Nyquist filter, such as the filter bank described in the MPEG Surround standard (see ISO/IEC document) can be used 23003-1). The complex QMF representation of the signal is oversampled by a factor of 2 because it is complex-valued and not real-valued. This allows time- and frequency-adaptive signal processing without audible aliasing artifacts. This hybrid filter bank typically provides high frequency resolution (narrow band) at low frequencies, while at high frequencies multiple QMF bands are grouped into a wider band. Article "Low Complexity Parametric Stereo Coding in MPEG-4", H. Purnhagen, Proc. of the 7th Int. Conference on Digital Audio Effects (DAFx'04), Naples, Italy, October 5-8, 2004, pages 163-168, describes an embodiment of a hybrid filter bank (see section 3.2 and Fig. 4). This disclosure is incorporated by reference herein. The specified document allows a sampling rate of 48 kHz with a (nominal) bandwidth from a 64-band QMF block of 375 Hz. The Bark perceptual frequency scale, however, requires a bandwidth of approximately 100 Hz for frequencies below 500 Hz. Therefore, the first 3 QMF bands can be split into even narrower subbands using a Nyquist filter bank. The first QMF band can be split into 4 bands (plus two more for negative frequencies), and the second and third QMF bands can be split into two bands each.

On the other hand, adaptive L/R or M/S encoding is preferably performed in the critically sampled MDCT domain (as described, for example, in AAC) in order to provide an efficient representation of the quantized signal. The conversion of the downmix signal DMX and the residual signal RES into the pseudo-stereo signal L _p , R _p in the conversion step 2 can be carried out in the time domain, since the PS encoder 1 and the perceptual encoder 3 can in any case be coupled in the time domain. In the decoding system, the perceptual stereo decoder 11 and the PS decoder 13 are also preferably coupled in the time domain. Therefore, the conversion of the pseudo-stereo signal L _p , R _p to the downmix signal DMX and the residual signal RES in the conversion step 12 can also be carried out in the time domain.

An adaptive stereo L/R or M/S encoder such as encoder 3 in FIG. 1 is typically a perceptual audio encoder that incorporates a psychoacoustic model that allows high efficiency encoding at low bit rates. An example of such an encoder is an AAC encoder, which uses critically sampled MDCT domain transform encoding in combination with time- and frequency-dependent quantization that is controlled using a psychoacoustic model. In addition, time- and frequency-dependent decision making between L/R and M/S encoding is typically guided by perceptual entropy measures that are computed using a psychoacoustic model.

Perceptual stereophonic an encoder (such as encoder 3 in Fig. 1) performs operations on the pseudo-stereo L/R signal (see Fig.L _p ,R _p in fig. 1). To optimize the coding efficiency of a stereo encoder (in particular, so that it makes the correct decision between L/R encoding and M/S encoding), it is preferable to modify the psychoacoustic control mechanism (including the control mechanism that makes the choice between stereo L /R-coding and stereo M/S-coding, a control mechanism that controls time- and frequency-dependent quantization) in the perceptual stereo encoder so that it takes into account signal modifications (pseudo-L/R conversion toDMX AndRES followed by PS encoding), which are used in the decoder to generate the final stereo output signal L, R. These signal modifications can affect the binaural masking effects that are exploited by psychoacoustic control mechanisms. Therefore, psychoacoustic control mechanisms should preferably be suitably adapted. To achieve this, it may be useful for the psychoacoustic control mechanisms to have access to more than just the pseudo-L/R signal (see Fig.L _p ,R _p in fig. 1), but also to the PS encoding parameters (see position 5 in Fig. 1) and/or to the original stereo signal L, R. The access of psychoacoustic control mechanisms to the PS encoding parameters and to the stereo signal L, R is shown in Fig. . 1 dashed line. For example, based on this information, the masking threshold(s) may be adapted.

An alternative approach to optimizing psychoacoustic control is to add to the encoder system a detector forming a deactivation step that has the ability to effectively deactivate PS encoding when necessary, preferably in a time- and frequency-dependent manner. Disabling PS encoding is appropriate, for example, when stereo L/R encoding is expected to be more useful, or when psychoacoustic control may have difficulty effectively encoding the pseudo-L/R signal. PS coding can be effectively disabled by specifying the downmix matrix H ^–1 such that the downmix H ^–1 matrix followed by the transform (see step 2 of FIG. 1) corresponds to an identity matrix (i.e., identification operation) or the identity matrix multiplied by a certain coefficient. For example, PS encoding can be effectively disabled by forcing the PS encoding parameters to be IID = 0 dB and ICC = 0. In this case, as discussed above, the pseudo-stereo signal L _p , R _p corresponds to the stereo signal L , R .

This PS encoding parameter modification control by the detector is shown in FIG. 4. In this case, the detector 20 receives the PS encoding parameters 5 determined by the parameter estimation step 9. When the detector does not deactivate the PS encoding, the detector 20 passes the PS encoding parameters to the downmixing step 8 and to the multiplexer 7, i.e. in this case, the PS encoding parameters 5 correspond to the PS encoding parameters 5' supplied to the downmixing step 8. In the event that the detector detects that PS encoding is unfavorable and should be deactivated (for one or more frequency bands), the detector modifies the affected PS encoding parameters 5 (eg, assigns the PS encoding parameters, IID and/or ICC, IID values = 0 dB and ICC = 0) and supplies the modified 5' PS encoding parameters to the downmix stage 8. The detector may also optionally consider the left and right L, R signals to decide whether to modify the PS encoding parameters (see dashed lines in FIG. 4).

In the following figures, the term QMF (quadrature mirror filter or filter bank) also includes a QMF subband filter bank in combination with a Nyquist filter bank, i.e. design of a hybrid filter block. In addition, all quantities in the following description may be frequency dependent, for example, different downmix and upmix matrices may be derived for different frequency ranges. In addition, the encoding of the residual signal may only cover a portion of the usable frequency range of the audio signal (ie, the residual signal is encoded for only a portion of the usable frequency range of the audio signal). The downmixing aspects described below for some frequency ranges may occur in the QMF domain (eg, in accordance with the prior art), while for other frequency ranges only, for example, phase aspects will occur in the complex QMF domain, and the conversion amplitudes will be produced in the real-valued MDCT region.

In fig. 5 shows a traditional PS encoder system. Each of the stereo channels L, R is first analyzed by a complex QMF 30 with M subbands, for example, QMF with M=64 subbands. The subband signals are used in the PS encoder 31 to estimate the PS encoding parameters 5 and the DMX downmix signal. The downmix DMX signal is used in the SBR encoder 32 to estimate the SBR (spectral band replication) parameters 33. SBR encoder 32 extracts SBR parameters 33 representing the spectral envelope of the original high-bandwidth signal, possibly in combination with noise and tonality criteria. Unlike PS encoder 31, SBR encoder 32 has no effect on the signal passing to base encoder 34. The DMX downmix signal of PS encoder 31 is synthesized using inverse QMF 35 with N subbands. For example, a complex QMF with N=32 can be used, where only the 32 lowest subbands out of the 64 subbands used by the PS encoder 31 and the SBR encoder 32 are synthesized. Thus, using half of the total number of subbands for the same frame size, the result is passes to base encoder 34 a time domain signal containing half the bandwidth of the input signal. Due to the reduced bandwidth, the sampling rate may be reduced by half (not shown). The base encoder 34 converts the perceptual encoding of the monaural input signal to generate a bitstream 36. PS encoding parameters are embedded into the bitstream 36 using a multiplexer (not shown).

In fig. 6 shows the following embodiment of an encoder system that combines PS encoding using a residual signal with a stereo base encoder 48, where the stereo base encoder 48 is suitable for adaptive perceptual stereo L/R and M/S encoding. This embodiment of the invention is merely an illustration of the principles of the present application. It should be understood that modifications and changes to this embodiment of the invention will be apparent to those skilled in the art. The input channels L, R, representing the original left and right channels, are analyzed by the complex QMF 30 in a manner similar to that discussed in connection with FIG. 5. Unlike the PS encoder 31 of FIG. 5, PS encoder 41 of FIG. 6 not only outputs the downmix DMX signal, but also outputs the residual RES signal. The DMX downmix signal is used by the SBR encoder 32 to determine the DMX downmix signal SBR parameters 33 . In conversion step 2, a fixed DMX / RES to pseudo-L/R conversion (eg M/S to L/R conversion) is applied to the DMX downmix and residual RES signals. Conversion stage 2 in FIG. 6 corresponds to transformation step 2 in FIG. 1. Conversion step 2 creates a "pseudo-" left and right channel signal L _p , R _p for operation by the base encoder 48. In this embodiment of the invention, the inverse L/R to M/S conversion is applied in the QMF domain before subband synthesis filter banks 35. Preferably, the number N (eg, N=32) of subbands for synthesis corresponds to half the number of M (eg, M=64) subbands used for analysis, and the base encoder 48 operates at half the sampling rate. It should be noted that there is no limitation to using 64 subband channels for QMF analysis in the encoder and 32 subbands for synthesis, other values are also possible depending on what sampling rate is desired for the signal received by the basic encoder 48. Basic Stereo Encoder 48 performs perceptual encoding of the signal of the filter banks 35 to generate a bit stream signal 46. The PS encoding parameters 5 are embedded into the 46 bitstream signal via a multiplexer (not shown). Optionally, the base encoder 48 may use the PS encoding parameters and/or the original L/R input signal. This information indicates to base encoder 48 how PS encoder 41 rotates in stereo space. This information can guide the base encoder 48 on how to control the quantization in a perceptually optimal manner. This is shown in Fig. 6 dashed lines.

In fig. 7 shows another embodiment of an encoder system that is similar to the system of FIG. 6. Unlike the embodiment of the invention shown in FIG. 6, Fig. 7, the SBR encoder 42 is connected upstream of the PS encoder 41. In FIG. 7, the SBR encoder 42 is placed before the PS encoder 41 and thus performs operations on the left and right channels (in this case, the QMF domain) instead of, as in FIG. 6, Perform operations on the downmixed DMX signal.

Due to the repositioning of SBR encoder 42, PS encoder 41 may be configured to perform operations not over the entire bandwidth of the input signal, but, for example, only over a frequency range below the SBR transition frequency. In fig. 7, the SBR parameters 43 for the SBR band are in stereo space, and the output of the corresponding PS decoder, as will be discussed later in connection with FIG. 15 generates the original stereo frequency range for operation by the SBR decoder. This modification, i.e. attaching the SBR encoder module 42 in an upstream direction relative to the PS encoder module 41 in the encoder system, and accordingly placing the SBR decoder module after the PS decoder in the decoder system (see FIG. 15), has the advantage that , which can reduce the use of a decorrelated signal to generate a stereo output signal. It should be noted that if there is no residual signal at all or none for a particular frequency band, a decorrelated version of the downmixed DMX signal is used in the PS decoder instead. However, reconstruction based on a decorrelated signal reduces the sound quality. Therefore, reducing the use of decorrelated signal results in increased audio quality.

This advantage of the embodiment of the invention according to FIG. 7 compared with the embodiment of the invention in FIG. 6 will be described in more detail with reference to FIG. 8a–8d.

In fig. 8a visualizes the time-frequency representation of one of the two output channels L, R (on the decoder side). In the case of FIG. 8a uses an encoder where the PS encoding module is placed before the SBR encoding module, as in the encoder of FIG. 5 or fig. 6 (in the decoder, the PS decoder is placed after the SBR decoder, see Fig. 14). In addition, the residual signal is encoded only in the low-bandwidth frequency range 50, which is smaller than the frequency range 51 of the base encoder. As can be seen from the visualization in Fig. 8a, frequency range 52, where the decorrelated signal to be used by the PS decoder covers the entire frequency range except the less high frequency range 50 covered by using the residual signal. In addition, SBR covers a frequency range of 53, which starts significantly higher than the frequency range of the decorrelated signal. The full frequency range is thus divided into the following frequency ranges: the lower frequency range (see band 50 in Fig. 8a) uses waveform encoding; in the mid-frequency range (see the area where frequency ranges 51 and 52 intersect) waveform coding is used in combination with a decorrelated signal; and in the higher frequency range (see frequency range 53) a regenerated SBR signal regenerated from lower frequencies is used in combination with a decorrelated signal that is generated by the PS decoder.

In fig. 8b, the time-frequency representation of one of the two output channels L, R (on the decoder side) is visualized for the case where the SBR encoder is connected upstream of the PS encoder in the encoder system (and the SBR decoder is located after the PS decoder in the decoder system ). In fig. 8b shows a low bit rate scenario where the residual signal bandwidth 60 (where the residual signal encoding is performed) is below that of the base encoder 61. Because the SBR decoding process is performed on the decoder side after the PS decoder (see FIG. 15), the residual signal used for the low frequencies is also used to reconstruct at least a portion (see frequency band 64) of the higher frequencies in the SBR band 63.

This advantage becomes even more apparent when the operation is performed at intermediate bit rates, where the bandwidth of the residual signal approaches, or becomes equal to, the bandwidth of the underlying encoder. In this case, the time-frequency representation of FIG. 8a (using the PS encoding and SBR encoding order of FIG. 6) results in the time-frequency representation shown in FIG. 8s. In fig. 8c, the residual signal covers substantially the entire low-band range 51 of the base encoder; In the frequency range 53 SBR, the PS decoder uses a decorrelated signal. In fig. 8d visualizes the time-frequency representation in the case of a preferred order of encoder/decoder modules (i.e., SBR encoding is performed on the stereo signal before PS encoding, as shown in FIG. 7). Here, the PS decoding unit in the decoder performs operations before the SBR decoding unit, as shown in FIG. 15. Thus, the residual signal is part of the low-frequency band used for high-frequency reconstruction. When the bandwidth of the residual signal becomes equal to the bandwidth of the downmixed monaural signal, no decorrelated signal information is required for the output signal decoder (see the full frequency range, which is shaded in Fig. 8d).

In fig. 9a shows an embodiment of a basic stereo encoder 48 with adaptive selection of stereo L/R or M/S encoding in the MDCT transform domain. Said stereo encoder 48 may be used in FIG. 6 and 7. The basic mono encoder 34 shown in FIG. 5 can be viewed as a special case of the basic stereo encoder 48 of FIG. 9a, in which only one mono input channel is processed (i.e. when there is no second input channel, shown in Fig. 9a with a dashed line).

In fig. Figure 9b shows a more general encoder. For monaural signals, the coding may switch between linear prediction domain coding (see Block 71) and transform domain coding (see Block 48). This type of encoder introduces multiple encoding methods that can be used adaptively depending on the characteristics of the input signal. In this case, the encoder can choose between encoding the signal using AAC-type encoder 48 (suitable for mono and stereo signals, with adaptive selection of L/R or M/S encoding in the case of stereo signals) and using encoder 71 , belonging to the AMR-WB+ (Adaptive Multi Rate - WideBand Plus) type, suitable only for monaural signals. The AMR-WB+ core encoder 71 evaluates the linear predictor residual signal 72 and in turn also makes a choice between a linear prediction residual signal transform encoding approach and a classic ACELP (Algebraic Code Excited Linear Prediction) speech encoder approach for encoding the linear prediction residual signal. To select between the AAC-type transform encoder 48 and the AMR-WB+ type base encoder 71, a mode selection decision block 73 is used, which decides between encoders 48 and 71 based on the input signal.

Encoder 48 is an MDCT-based AAC-type encoder. If the mode decision step 73 directs the input signal to use MDCT-based encoding, the signals input monaural input signal or stereo input signals are encoded by the AAC-based MDCT encoder 48. MDCT encoder 48 performs MDCT analysis of one or two signals in MDCT steps 74. In the case of a stereo signal, step 75 also makes an M/S or L/R selection decision based on the frequency band before quantizing and encoding. Stereo L/R encoding or stereo M/S encoding is selected in a frequency-dependent manner. Step 75 also performs L/R to M/S conversion. If an M/S coding decision is made for a particular frequency band, step 75 outputs an M/S signal for that frequency band. Otherwise, step 75 outputs an L/R signal for that frequency band.

Thus, if a transform coding mode is used, the stereo coding function of the underlying base encoder can be used for the stereo signal with full efficiency.

If the mode decision step 73 sends a mono signal to the encoder 71 in the linear prediction domain, the mono signal is then analyzed by linear prediction analysis in block 72. A decision is then made to encode the residual signal LP by an ACELP-type encoder 76 in the time domain or by the encoder 77, belonging to the TCX (Transform Coded eXcitation) type, which performs operations in the MDCT area. Linear prediction encoder 71 does not have any inherent stereo coding capability. Therefore, in order to allow the linear prediction domain encoder 71 to encode a stereo signal, an encoder configuration similar to that shown in FIG. 5. In this configuration, the PS encoder generates PS encoding parameters 5 and a downmixed mono DMX signal, which is then encoded by the encoder in the linear prediction domain.

In fig. 10 shows the following embodiment of an encoder system, where the parts of FIG. 7 and fig. 9 combined in a new way. DMX / RES to pseudo-L/R conversion block 2 described in FIG. 7 is located within an AAC downcoder 70 prior to the stereo MDCT analysis step 74. This embodiment has the advantage that the 2 DMX / RES to pseudo-L/R conversion is only applied when the underlying stereo MDCT encoder is used. Therefore, when the transform coding mode is used, the full efficiency of the stereo coding function of the underlying base encoder can be used to stereo code the frequency range covered by the residual signal.

While the mode selection decision step 73 of FIG. 9b performs operations on both the mono input signal and the stereo input signal, decision step 73' of FIG. 10 performs operations on the downmix DMX signal and the residual RES signal. In the case of a mono input signal, the mono signal can be directly used as a DMX signal, the RES signal can be set to zero, and the default PS encoding parameters can be set to IID = 0 dB and ICC = 1.

If the mode decision step 73' sends the downmixed DMX signal to the linear prediction domain encoder 71, the downmixed DMX signal is then analyzed by linear prediction analysis in block 72. A decision is then made as to whether to encode the residual LP signal by the ACELP-encoder 76. type in the time domain or through an encoder 77 of the TCX (Transform Coded eXcitation) type, which performs operations in the MDCT domain. Linear prediction encoder 71 does not have any inherent stereo coding capability that could be used to encode a residual signal in addition to the downmix DMX signal. Therefore, when the DMX downmix signal is encoded by the encoder 71 in the linear prediction domain, a dedicated residual signal encoder 78 is used to encode the residual signal RES . This encoder may, for example, be a monaural AAC encoder.

It should be noted that encoders 71 and 78 of FIG. 10 may be skipped (in which case the mode selection decision step 73' is no longer necessary).

In fig. 11a shows details of a further alternative embodiment of an encoder system that achieves the same benefits as the embodiment of FIG. 10. Unlike the embodiment of FIG. 10, in fig. 11a, the DMX / RES to pseudo-L/R conversion 2 is placed after the MDCT analysis 74 of the base encoder 70, i.e. the conversion is performed in the MDCT region. The transformation in block 2 is linear and time independent and can therefore be placed after the MDCT analysis 74. The remaining blocks of FIG. 10, which are not shown in FIG. 11 may optionally be added in the same way as in FIG. 11a. Alternatively, MDCT analysis blocks 74 may also be placed after transform block 2.

In fig. 11b illustrates one implementation of the embodiment of FIG. 11a. In fig. 11b shows an exemplary implementation of step 75 of selecting between M/S encoding and L/R encoding. Step 75 includes a sum and difference conversion step 98 (more precisely, an L/R to M/S conversion step), which receives the pseudo-stereo signal L _p , R _p . Conversion step 98 generates a pseudo-mid/side signal M _p , S _p by performing an L/R to M/S conversion. With the exception of possible gain, the following equations apply: M _p = DMX , S _p = RES .

Step 75 decides between L/R encoding and M/S encoding. Based on the selection made, either a pseudo-stereo signal L _p , R _p , or a pseudo-mid/side signal M _p , Sp _, is selected and encoded in AAC block 97 . It should be noted that two AAC blocks 97 (not shown in FIG. 11b) may also be used, where the first AAC block 97 is for the pseudo-stereo signal L _p , R _p , and the second AAC block 97 is for the pseudo mid/side signal M _p , S _p . In this case, the selection of L/R or M/S is made by selecting either the output signal of the first AAC block 97 or the output signal of the second AAC block 97.

In fig. 11c shows an alternative embodiment of the invention of FIG. 11a. In this case, stage 2 of the transformation is not used explicitly. Instead, transform step 2 and step 75 are combined into a single step 75'. The downmix DMX signal and the residual RES signal are supplied to a sum and difference conversion step 99 (more precisely, a DMX / RES to pseudo-L/R conversion step) forming part of a step 75'. Conversion step 99 generates a pseudo-stereo signal L _p , R _p . DMX / RES to pseudo-L/R conversion step 99 of FIG. 11c is similar to the L/R to M/S conversion step 98 of FIG. 11b (except possibly for different gain). However, in FIG. 11c, the choice between M/S decoding and L/R decoding must be inverted with respect to FIG. 11b. It should be noted that in FIG. 11b and 11c, the switch position for selecting L/R or M/S is shown in the L _p / R _p position, which is the top position in FIG. 11b and lower - in Fig. 11s. This visualizes the inverse meaning of choosing L/R or M/S.

It should be noted that in FIG. 11b and 11c, there is preferably a separate switch for each frequency band in the MDCT region, and thus the choice between L/R and M/S can be both time dependent and frequency dependent. In other words, the switch position preferably depends on the frequency. Conversion steps 98 and 99 may convert the entire usable frequency range or just one frequency band.

Additionally, note that blocks 2, 98, and 99 can all be called "sum-difference conversion blocks" because all of these blocks implement a conversion matrix of the form

.

The only thing that may differ in blocks 2, 98, 99 is the gain c .

In fig. 12 describes the following embodiment of an encoder system. It uses an extended set of PS encoding parameters, which, in addition to IID and ICC (described above), includes two more parameters: IPD (phase difference between channels, see below). below) and OPD (overall phase difference, see below), which allow one to characterize the phase relationship between two channels, L and R, of a stereo signal. An example of these phase parameters is given in ISO/IEC 14496-3, clause 8.6.4.6.3, which is incorporated by reference herein. When using phase parameters, the resulting matrix upmixing (and its inverse matrix ) becomes complex-valued in accordance with the equation:

,

Where

,

and where

.

PS encoder stage 80, which performs operations in the complex QMF domain, takes into account only the phase dependencies of the L, R channels. Downward rotation (i.e., the conversion from the L/R domain to the DMX / RES domain, which is described by the above matrix H ^–1 ) is taken into account in the MDCT domain forming part of the basic stereo encoder 81. Therefore, the phase relationships between the two channels are extracted in the complex QMF domain, while the remaining real-valued waveform relationships are extracted in the real-valued domain of the critically sampled MDCT forming part the stereo encoding mechanism of the underlying encoder used. This has the advantage that the extraction of linear relationships between channels can be tightly integrated into the stereo encoding of the underlying encoder (although, to prevent aliasing, in the critically sampled MDCT region, only for the frequency range that is covered by the residual signal encoding, possibly minus "guard band" on the frequency axis).

PS encoder phase adjustment step 80 of FIG. 12 extracts phase-related PS encoding parameters, such as IPD (inter-channel phase difference) and OPD (overall phase difference) parameters. Therefore the matrix $$H_{\varphi }^{-1}$$ The regulation of the phase it generates can correspond to the following equation:

.

As discussed above, the downward rotation portion of the PS encoding unit deals with the stereo encoding unit 81 of the base encoder of FIG. 12. A stereo encoding unit 81 that performs operations in the MDCT region is shown in FIG. 13. Stereo encoding unit 81 receives stereo signal with phase adjusted MDCT region. This signal is downmixed in a matrix downmix step 82 downward rotation, which, as discussed above, is the real-valued part of the complex matrix downmix, and thus the downmix DMX signal and the residual signal RES are generated. The downmix operation is followed by L/R to M/S conversion according to the present application (see conversion step 2), and thus a pseudo-stereo signal L _p , R _p is generated. The pseudo-stereo signal L _p , R _p is processed by a stereo encoding algorithm (see adaptive stereo M/S or L/R encoder 83), in this particular embodiment of the invention the stereo encoding mechanism, which depends on perceptual entropy criteria, adopts decision to encode the L/R representation of the signal or the M/S representation of the signal. This decision is preferably time and frequency dependent.

In fig. 14 shows an embodiment of a decoder system that is suitable for decoding a bitstream 46 generated by the encoder system shown in FIG. 6. This embodiment of the invention is merely an illustration of the principles of the present application. It should be understood that modifications and changes to this embodiment of the invention will be apparent to those skilled in the art. The base decoder 90 decodes the bitstream 46 into pseudo-left and pseudo-right channels, which are converted into the QMF domain by filter banks 91. Conversion step 12 then converts the resulting pseudo-stereo signal L _p , R _p from L/R to DMX / RES , and thus the DMX downmix signal and the residual signal RES are created. When using SBR encoding, these signals are low-bandwidth signals, for example, the downmix DMX signal and the residual RES signal may only contain audio signal information for the low-bandwidth frequency up to approximately 8 KHz. The downmix DMX signal is used by the SBR decoder 93 to reconstruct the high frequency band based on the received SBR parameters (not shown). The output of the SBR decoder 93 (including the low-pass and reconstructed high-pass bands of the DMX downmix signal) and the residual signal RES are input to the PS decoder 94, which performs operations in the QMF domain (specifically, in the domain of the hybrid QMF filter + Nyquist filter block). The downmix signal DMX at the input of the PS decoder 94 also contains information about the audio signal in the high frequency band (for example, up to 20 KHz), while the residual signal RES at the input to the PS decoder 94 is a low band signal (for example, limited frequency up to 8 kHz). Therefore, for the high frequency band (eg, the 8 KHz to 20 KHz band), PS decoder 94 uses a decorrelated version of the downmixed DMX signal instead of the band-limited residual signal RES . Thus, the decoded signals at the output of PS decoder 94 are based on the residual signal only up to 8 KHz. After PS decoding, both output channels of the PS decoder 94 are converted to the time domain by the filter banks 95, and thus a stereo output signal L, R is generated.

In fig. 15 shows an embodiment of a decoding system that is suitable for decoding a bit stream 46 generated by the encoder system of FIG. 7. This embodiment of the invention is solely an illustration of the principles of the present application. It should be understood that modifications and changes to this embodiment of the invention will be apparent to those skilled in the art. The operating principle of the embodiment of the invention shown in FIG. 15 is similar in principle to the decoder system described in FIG. 14. Unlike FIG. 14, SBR decoder 96 in FIG. 15 is located at the output of PS decoder 14. In addition, the SBR decoder uses SBR parameters (shown) to generate spectral envelope data, as opposed to the mono SBR parameters of FIG. 14. The downmix and residual signals at the input of the PS decoder 94 are typically low-band signals, for example, the DMX downmix signal and the RES residual signal may contain audio signal information only for the low-frequency band, for example, with a frequency of up to approximately 8 KHz. Based on the low-bandwidth downmix DMX signal and the residual RES signal, PS encoder 94 determines a low-bandwidth stereo signal, for example, with a frequency of up to approximately 8 KHz. Based on the low-bandwidth stereo signal and the SBR stereo parameters, the SBR decoder 96 reconstructs the high-frequency portion of the stereo signal. Compared with the embodiment of the invention in FIG. 14, embodiment of the invention according to FIG. 15 provides the advantage that a decorrelated signal is not required (see also FIG. 8d) and thus improved sound quality is achieved, while in FIG. 14, the high-frequency part requires a decorrelated signal (see also FIG. 8c), which thus reduces the sound quality.

In fig. 16a shows an embodiment of a decoding system which is the inverse of the encoding system of FIG. 11a. The input bit stream signal is supplied to a decoder block 100, which generates a first decoded signal 102 and a second decoded signal 103. At the decoder, M/S encoding or L/R encoding is selected. The choice is indicated in the received bitstream. Based on this information, a selection of M/S or L/R is made at step 101. If M/S is selected by the decoder, the first 102 and second 103 signals are converted to a (pseudo-) L/R signal. In case L/R is selected by the decoder, the first 102 and second 103 signals may pass through step 103 without conversion. The pseudo-L/R signal, L _p , R _p , at the output of step 101 is converted into a DMX / RES signal through conversion step 12 (this step performs a quasi-L/R to M/S conversion). Preferably, steps 100, 101 and 12 of FIG. 16a perform operations in the MDCT region. Conversion blocks 104 may be used to convert the downmixed DMX signal and the residual RES signal into the time domain. The resulting signal is then fed to a PS decoder (not shown) and optionally to an SBR decoder, as shown in FIG. 14 and 15. Alternatively, blocks 104 may also be placed in front of block 12.

In fig. 16b illustrates an implementation of the embodiment of FIG. 16a. In fig. 16b shows an exemplary implementation of step 101 for selecting between M/S decoding and L/R decoding. Step 101 includes a sum and difference conversion (M/S to L/R conversion) step 105 that receives the first 102 and second 103 signals.

Based on the encoding information provided in the bit stream, step 101 makes a choice between L/R encoding and M/S encoding. If the choice is made in favor of L/R encoding, then the output signal of the decoding unit 100 is supplied to the conversion stage 12.

In fig. 16c shows an alternative to the embodiment of FIG. 16a. In this case, conversion step 12 is not used explicitly. Instead, the conversion step 12 and step 101 are combined into a single step 101'. The first 102 and second 103 signals are fed to a sum and difference conversion step 105' (more precisely, a pseudo-L/R to DMX / RES conversion step) forming part of step 101'. The conversion step 105' generates the DMX / RES signal. The conversion step 105' of FIG. 16c is similar to the conversion step 105 of FIG. 16b or identical to it (except possibly with a different gain). In fig. 16c, the choice between M/S decoding and L/R decoding must be inverted with respect to FIG. 16b. In fig. 16c the switch is in the down position, while in FIG. 16b switch is in the up position. This is how the inversion of the L/R or M/S selection is visualized (the selection signal can be inverted simply by means of an inverter).

It should be noted that in FIG. 16b and 16c, there is preferably an individual switch for each frequency band in the MDCT region, and thus the choice between L/R and M/S can be both time and frequency dependent. Conversion steps 105 and 105′ may convert the entire usable frequency range or only a single frequency band.

In fig. 17 shows the following embodiment of an encoding system for encoding a stereo signal L, R into a bitstream signal. The encoding system includes a downmixing step 8 for generating, based on the stereo signal, a downmix DMX signal and a residual signal RES . In addition, the encoding system includes a parameter determination step 9 for determining one or more parametric stereo encoding parameters 5. In addition, the encoding system includes perceptual encoding means 110 downstream of the downmixing step 8. The following encoding modes are selectable:

- encoding, based on the sum signal of the DMX downmix signal and the residual RES signal and on the difference signal of the DMX downmix signal and the residual RES signal, or

- coding based on the downmixed DMX signal and the residual RES signal.

Preferably, the choice depends on time and frequency.

The encoding means 110 includes a sum and difference conversion step 111 that generates the sum and difference signals. In addition, the encoding means 110 includes a selector 112 for selecting encoding based on the sum and difference signals or based on the downmix signal DMX and the residual signal RES . In addition, a coding block 113 is provided. Alternatively, two encoding blocks 113 may be used, where the first encoding block 113 encodes the DMX and RES signals, and the second encoding block 113 encodes the sum and difference signals. In this case, the selection block 112 is located in the downstream direction relative to the two encoding blocks 113.

The sum and difference conversion block 111 has the form:

.

The conversion unit 111 may correspond to the conversion unit 99 of FIG. 11s.

The output of the perceptual encoder 110 is combined with the parameters 5 of the parametric stereo signal in the multiplexer 7, forming the resulting bit stream 6.

Unlike the design of FIG. 17, encoding based on the DMX downmix signal and the RES residual signal can be realized by encoding a resultant signal that is generated by converting the DMX downmix signal and the RES residual signal through two successive sum and difference conversions, as shown in FIG. 11b (see two conversion blocks 2 and 98). The resulting signal after two sum and difference conversions corresponds to the downmixed DMX signal and the residual RES signal (except for a possibly different gain).

In fig. 18 shows an embodiment of a decoder system that is the inverse of the encoder system of FIG. 17. The decoder system includes means 120 for perceptual decoding based on the bitstream signal. Before decoding, the PS encoding parameters are separated from the bitstream signal 6 in the demultiplexer 10. The decoding means 120 includes a base decoder 121, which (by decoding) generates a first signal 122 and a second signal 123. The decoding means outputs the downmix signal DMX and the residual signal RES .

The downmix DMX signal and the residual RES signal are selective

- based on the sum of the first signal 122 and the second signal 123 and on the difference of the first signal 122 and the second signal 123,

or

- based on the first signal 122 and the second signal 123.

Preferably, the choice depends on time and frequency. The selection is made at selection step 125.

The decoding means 120 includes a sum and difference step 124 that generates the sum and difference signals.

The conversion to sum and difference in block 124 is of the form:

.

Transformation block 124 may correspond to block 105' of FIG. 16s.

Once selected, the DMX and RES signals are supplied to an upmixing step 126 to generate a stereo L, R signal based on the downmix DMX signal and the residual RES signal. The upmixing operation depends on the PS encoding parameters 5.

Preferably, the selection in FIG. 17 and 18 is frequency dependent. In fig. 17, for example, a time-to-frequency conversion (eg, via an MDCT or analysis filter bank) may be performed as a first step in the perceptual encoding means 110. In fig. 18, for example, frequency-to-time conversion (eg, via an inverse MDCT or synthesis filter bank) may be performed as a final step in the perceptual decoding means 120.

It should be noted that in the embodiments described above, the signals, parameters and matrices may be frequency dependent or frequency independent, and/or time dependent or time independent. The calculation steps described can be performed for individual frequencies or for the entire frequency band of the audio signal.

Additionally, it should be noted that all the various sum and difference conversions, i.e. DMX / RES to pseudo-L/R conversion, pseudo-L/R to DMX / RES conversion, L/R to M/S conversion, and M/S to L/R conversion are of the form

.

The only thing that may differ is the gain c . Therefore, in principle, each of these transformations can be replaced by another transformation from among the specified transformations. If the gain during the encoding process is not correct, it can be compensated during the decoding process. Also, if you place two identical or two different sum and difference transformations in series, the resulting transformation will correspond to an identity matrix (possibly multiplied by a gain).

In an encoder system including a PS encoder and an SBR encoder, various PS/SBR configurations are possible. In the first configuration shown in FIG. 6, the SBR encoder 32 is connected downstream of the PS encoder 41. In the second configuration shown in FIG. 7, an SBR encoder 42 is coupled upstream of a PS encoder 41. Depending on, for example, the desired target bit rate, properties of the underlying encoder, and/or one or more different factors, one configuration may be preferred over other configurations. In general, for lower bit rates, the first configuration may be more preferable, while for higher bit rates, the second configuration may be more preferable. It is therefore desirable for the encoder system to support both of these configurations so that it is capable of making a selection of the preferred configuration depending on, for example, the desired target bit rate and/or one or more other criteria.

In a decoder system including a PS decoder and an SBR decoder, various PS/SBR configurations are also possible. In the first configuration shown in FIG. 14, the SBR decoder 93 is connected upstream of the PS decoder 94. In the second configuration shown in FIG. 15, the SBR decoder 96 is connected downstream of the PS decoder 94. To achieve correct operation, the decoder system configuration must match the encoder system configuration. If the encoder is configured according to FIG. 6, the decoder is accordingly configured in accordance with FIG. 14. If the encoder is configured according to FIG. 7, the decoder is accordingly configured in accordance with FIG. 15. To ensure correct operation, the encoder preferably provides a signal to the decoder about the PS/SBR configuration that has been selected for encoding (and thus the PS/SBR configuration that should be selected for decoding). Based on this information, the decoder makes a selection of the appropriate decoder configuration.

As discussed above, to ensure correct operation of the decoder, there is preferably a mechanism for transmitting from the encoder to the decoder a signal about the configuration that should be used in the decoder. This may be done explicitly (eg, by a dedicated bit or field in the bitstream configuration header, as discussed below) or implicitly (eg, by checking whether the SBR data is mono or stereo when PS encoding data is present).

As discussed above, a dedicated element in the header of the bit stream passed from the encoder to the decoder can be used to signal the selected PS/SBR configuration. The specified bitstream header carries the necessary configuration information that is required to enable the decoder to correctly decode the data in the bitstream. A specialized element in the bitstream header may be, for example, a one-bit flag, field, or index pointing to a specific table entry that defines various decoder configurations.

Instead of including an additional specialized element in the bitstream header to signal the PS/SBR configuration, the decoding system can evaluate information already present in the bitstream to select the correct PS/SBR configuration. For example, the PS/SBR configuration selection may be inferred from the PS decoder and SBR decoder configuration information in the bitstream header. The configuration information typically indicates whether the SBR decoder should be configured for mono operations or for stereo operations. If, for example, the PS decoder is enabled and the SBR decoder is configured for mono operations (as indicated in the configuration information), the PS/SBR configuration of FIG. 14. If, for example, the PS decoder is enabled and the SBR decoder is configured for stereo operations (as indicated in the configuration information), the PS/SBR configuration may be selected as shown in FIG. 15.

The embodiments described above are merely illustrative of the principles of the present application. It should be understood that changes and modifications to the configurations and details described herein will be apparent to those skilled in the art. Therefore, it is not intended that the scope of the application be limited to the specific details provided by the description and explanation of embodiments of the invention in this specification.

The systems and methods disclosed herein may be implemented as software, firmware, hardware, or combinations thereof. Some or all of the components may be implemented as software running on a digital signal processor or microprocessor, or implemented as hardware or application specific integrated circuits.

Typical devices employing the disclosed systems and methods include portable audio players, mobile communications devices, set-top boxes, televisions, AVRs (audio video tuners), personal computers, etc.

Claims (79)

1. A method for encoding a stereo input signal comprising a left channel and a right channel and having a perceptual stereo image, the method comprising: selecting either a transform coding mode or a linear prediction coding mode as the selected coding mode; encoding the stereo input signal using only the selected encoding mode to obtain an encoded output signal; And generating a bitstream signal containing the encoded output signal, Moreover, if linear prediction coding mode is selected, coding includes: downmixing the stereo input signal into a mono signal, the mono signal being the sum of a left channel and a right channel, estimating stereo picture parameters, to reconstruct a stereo signal that approximates the perceptual stereo picture of a stereo input signal, from a mono signal, generating a residual signal that indicates the error associated with mapping a stereo signal to a mono signal, and the estimated parameters of the stereo picture, encoding the monaural signal using linear prediction coding to obtain the encoded monaural signal, and outputting the encoded monaural signal, the residual signal, and the stereo picture parameters as the encoded output signal, Moreover, if the transform encoding mode is selected, encoding includes: analyzing a stereo input signal by applying both mid/side stereo encoding and left/right stereo encoding, and selecting either a mid/side stereo encoding mode or a left/right stereo encoding mode based on the estimated entropy for each encoding mode stereo signal, encoding the stereo input signal using a selected stereo encoding mode in the first frequency range to obtain an encoded stereo signal in the first frequency range, downmixing a stereo input signal into a mono signal in a second frequency range, encoding the monaural signal in the second frequency range using transform coding to obtain an encoded monaural signal in the second frequency range; and outputting the encoded stereo signal in the first frequency range and the encoded monaural signal in the second frequency range as an encoded output signal. 2. The method of claim 1, wherein the analysis includes selecting which stereo encoding mode will more efficiently encode the stereo input signal. 3. The method according to claim 1, characterized in that the choice of either a transform coding mode or a linear prediction mode depends on the characteristics of the stereo input signal. 4. The method of claim 1, wherein the transform coding further includes not encoding the one or more subbands and generating additional information for reconstructing the one or more subbands. 5. The method of claim 4, wherein the additional information includes a parameter used to determine the spectral envelope of one or more subbands not subject to coding. 6. The method according to claim 1, characterized in that the transform coding includes a psychoacoustic model. 7. The method according to claim 1, characterized in that the evaluation includes estimating the parameters of the stereophonic picture in a variety of frequency subranges. 8. The method of claim 1, wherein the bandwidth of the first frequency range and the bandwidth of the second frequency range are determined based at least in part on the desired target bit rate. 9. The method of claim 1, wherein the linear prediction coding mode is selected when the stereo input signal is speech. 10. A non-volatile computer-readable medium containing instructions that, when executed by a processor, carry out the method according to claim 1. 11. An apparatus for encoding a stereo input signal comprising a left channel and a right channel and having a perceptual stereo image to obtain an encoded output signal, the apparatus comprising: a mode selector for selecting either a transform coding mode or a linear prediction coding mode; a transform encoder for encoding a stereo input signal if the selected encoding mode is a transform encoding mode, but not if the selected encoding mode is a linear prediction encoding mode; a linear prediction encoder for encoding a stereo input signal if the selected encoding mode is a linear prediction encoding mode, but not if the selected encoding mode is a transform encoding mode; And a bitstream generator for generating a bitstream signal comprising a coded output signal, wherein the linear prediction encoder is configured to: downmixing the stereo input signal into a mono signal, wherein the mono signal is the sum of the left channel and the right channel, estimating stereo picture parameters, to reconstruct a stereo signal that approximates the perceptual stereo picture of a stereo input signal, from a mono signal, generating a residual signal that indicates the error associated with the mapping of a stereo signal to a mono signal, and the estimated parameters of the stereo picture, encoding the monaural signal using linear prediction coding to obtain the encoded first monaural signal and output the encoded monaural signal, the residual signal, and the estimated stereo picture parameters as the encoded output signal, wherein the encoder with transformation is configured to: analyzing the stereo input signal by applying both mid/side stereo encoding and left/right stereo encoding, and selecting either a mid/side stereo encoding mode or a left/right stereo encoding mode based on the estimated entropy for each encoding mode stereo signal, encoding the stereo input signal using a selected stereo encoding mode in the first frequency range to obtain an encoded stereo signal in the first frequency range, downmixing a stereo input signal into a mono signal in a second frequency range, encoding a monaural signal in a second frequency range using transform coding to obtain an encoded monaural signal in a second frequency range; and outputting the encoded stereo signal in the first frequency range and the encoded monaural signal in the second frequency range as an encoded output signal. 12. A method for decoding a bit stream signal to obtain a decoded output signal having a left channel and a right channel, the method comprising: extracting an encoded audio signal from a bitstream signal, wherein the encoded audio signal is generated by encoding an input stereo audio signal having a left input channel and a right input channel using a selected encoding mode, wherein the selected encoding mode is one of a transform encoding mode or a linear prediction coding; decoding the encoded audio signal using only the selected encoding mode to obtain a decoded signal; And output the decoded signal as a decoded output signal, wherein, if the selected encoding mode is a linear prediction encoding mode, decoding includes: receiving a coded mono signal, wherein the coded mono signal is the sum of a left input channel and a right input channel of the input stereo audio signal, decoding the encoded mono signal using linear prediction decoding to obtain the decoded mono signal, extracting stereo image parameters and a residual signal from the bitstream signal to reconstruct a stereo audio signal that approximates the perceptual stereo image of the input stereo audio signal, wherein the residual signal indicates the error associated with the mapping of the stereo audio signal to the monaural signal, and the stereo image parameters, reconstructing the stereo audio signal using the decoded mono signal, the residual signal, and the stereo image parameters to obtain a reconstructed stereo audio signal that approximates the perceptual stereo image of the input stereo audio signal, and output the reconstructed stereo audio signal as a decoded signal, however, if the selected encoding mode is a transform encoding mode, decoding includes: receiving a stereo signal in a first frequency range, wherein the stereo signal is generated using a selected stereo encoding mode, wherein the selected stereo encoding mode includes either mid/side stereo encoding or left/right stereo encoding, receiving a coded monophonic signal in the second frequency range, decoding the stereo signal in the first frequency range using the selected stereo encoding mode to obtain a decoded stereo signal in the first frequency range, decoding the encoded monaural signal in the second frequency range using transform decoding to obtain a decoded monaural signal in the second frequency range; and outputting the decoded stereo signal in the first frequency range and the decoded monaural signal in the second frequency range as a decoded signal. 13. The method of claim 12, wherein the transform coding also includes extracting additional information from the bitstream signal to reconstruct one or more subbands not subject to coding. 14. The method of claim 13, wherein the additional information includes a parameter used to determine the spectral envelope of one or more subbands not subject to coding. 15. The method according to claim 12, characterized in that the transform coding includes a psychoacoustic model. 16. The method according to claim 12, characterized in that the stereo picture parameters comprise stereo picture parameters for each of the plurality of frequency ranges. 17. The method of claim 12, wherein the bandwidth of the first frequency range and the bandwidth of the second frequency range are determined based at least in part on the desired target bit rate. 18. An apparatus for decoding a bitstream signal to obtain a decoded output signal having a left channel and a right channel, the apparatus comprising: a demultiplexer for extracting an encoded audio signal from a bitstream signal, wherein the encoded audio signal is generated by encoding an input stereo audio signal having a left input channel and a right input channel using a selected encoding mode, wherein the selected encoding mode is one of a transform encoding mode or linear prediction coding mode; a transform decoder for decoding the encoded audio signal if the selected encoding mode is a transform encoding mode, but not if the selected encoding mode is a linear prediction encoding mode; and a linear prediction decoder for decoding the encoded audio signal if the selected encoding mode is a linear prediction encoding mode, but not if the selected encoding mode is a transform encoding mode, wherein the linear prediction decoder is configured to: receiving a coded mono signal, wherein the coded mono signal is the sum of a left input channel and a right input channel of the input stereo audio signal, decoding the encoded mono signal using linear prediction decoding to obtain the decoded mono signal, extracting stereo image parameters and a residual signal from the bitstream signal to reconstruct a stereo audio signal that approximates the perceptual stereo image of the input stereo audio signal, wherein the residual signal indicates the error associated with the mapping of the stereo signal to the monaural signal, and the stereo image parameters, reconstructing the stereo audio signal using the decoded mono signal, the residual signal, and the stereo image parameters to obtain a reconstructed stereo audio signal that approximates the perceptual stereo image of the input stereo audio signal, and output the reconstructed stereo audio signal as a decoded output signal, wherein the decoder with conversion is configured to: receiving a stereo signal in a first frequency range, wherein the stereo signal is generated using a selected stereo encoding mode, wherein the selected stereo encoding mode includes either a mid/side stereo encoding mode or a left/right stereo encoding mode, receiving a coded monophonic signal in the second frequency range, decoding a stereo signal in the first frequency range using a selected stereo encoding mode to obtain a decoded stereo signal in the first frequency range, decoding the encoded mono signal in the second frequency range using transform decoding to obtain a decoded mono signal in the second frequency range; and outputting the decoded stereo signal in the first frequency range and the decoded monaural signal in the second frequency range as a decoded output signal. 19. The apparatus of claim 18, wherein the transform coding also includes extracting additional information from the bitstream signal to reconstruct one or more subbands not subject to coding. 20. The device of claim 19, wherein the additional information includes a parameter used to determine the spectral envelope of one or more subbands not subject to coding. 21. The device according to claim 18, characterized in that the transform coding includes a psychoacoustic model. 22. The device according to claim 18, characterized in that the stereo picture parameters contain parameters for each of a plurality of frequency ranges. 23. The apparatus of claim 18, wherein the bandwidth of the first frequency range and the bandwidth of the second frequency range are determined based at least in part on the desired target data rate.