EP1525576A1

EP1525576A1 - Arrangement and method for the generation of a complex spectral representation of a time-discrete signal

Info

Publication number: EP1525576A1
Application number: EP03766165A
Authority: EP
Inventors: Bernd Edler; Stefan Geyersberger
Original assignee: Fraunhofer Gesellschaft zur Forderung der Angewandten Forschung eV
Current assignee: Fraunhofer Gesellschaft zur Forderung der Angewandten Forschung eV
Priority date: 2002-07-26
Filing date: 2003-07-14
Publication date: 2005-04-27
Anticipated expiration: 2023-07-14
Also published as: WO2004013839A1; US20100161319A1; US7707030B2; US8155954B2; ATE432524T1; AU2003250945A1; EP1525576B1; DE50311552D1; DE10234130B3; US20050197831A1

Abstract

A filter bank device for generating a complex spectral representation of a discrete-time signal includes a generator for generating a block-wise real spectral representation, which, for example, implements an MDCT, to obtain temporally successive blocks of real spectral coefficients. The output values of this spectral conversion device are fed to a post-processor for post-processing the block-wise real spectral representation to obtain an approximated complex spectral representation having successive blocks, each block having a set of complex approximated spectral coefficients, wherein a complex approximated spectral coefficient can be represented by a first partial spectral coefficient and by a second partial spectral coefficient, wherein at least one of the first and second partial spectral coefficients is determined by combining at least two real spectral coefficients. A good approximation for a complex spectral representation of the discrete-time signal is obtained by combining two real spectral coefficients, preferably by a weighted linear combination, wherein additionally more degrees of freedom for optimizing the entire system are available.

Description

Device and method for generating a complex spectral representation of a discrete-time signal

description

The present invention relates to time-frequency conversion algorithms and in particular to such algorithms in connection with audio compression concepts.

For some applications in coding for data compression purposes and in particular in audio coding, it is necessary to display real-time discrete signals in the form of complex-valued spectral components. A complex special coefficient can be represented by a first and a second partial spectral coefficient, the first partial spectral coefficient being the real part and the second partial spectral coefficient being the imaginary part, as desired. Alternatively, the complex spectral coefficient can also be represented by the amount as the first partial spectral coefficient and the phase as the second partial spectral coefficient.

Real-valued transformation methods are often used in particular in audio coding, such as e.g. B. the well-known MDCT, which in "Analysis / Synthesis Filter Bank Design Based on Time Domain Aliasing Cancellation", J. Princen, A. Bradley, IEEE Trans. Acoust., Speech, and Signal Processing 34, p. 1153 - 1161, 1986. There is, for example, a need for a complex spectrum within the psychoacoustic model, for which reference is made to the psychoacoustic model in Annex D.2.4 of the ISO / IEC 11172-3 standard, which is also known as For certain applications, a complex discrete Fourier transformation runs in parallel with the actual MDCT transformation (MDCT = modified discrete cosine transformation) in order to calculate psychoacoustic parameters, such as the psychoacoustic masking threshold. With this discrete Fourier transform (DFT), the input signal is first divided into blocks of a predetermined length by means of multiplication with temporally offset window functions. Each of these blocks is then converted into a spectral representation using the DFT. If the blocks used each contain L sample values, ie if the window length is L, then the output of the DFT can again be completely described in the form of a total of L values (real and imaginary parts or magnitude and phase values). For example, if the input signal is real, L / 2 will result in complex values. When using suitable window functions, the input signal can be reconstructed from this representation using an inverse DFT.

However, there are some limitations to this approach. For example, critical scanning is only possible if successive ¹ windows do not overlap. Otherwise, with a temporal offset of N <L values for each N new input values of the DFT L values in the spectral representation would have to be transmitted, which is particularly undesirable in data compression methods.

However, the use of non-overlapping window functions severely limits the quality of the spectral decomposition that can be achieved, the separation of different frequency bands being particularly noteworthy.

On the other hand, a better band separation can be achieved with real-valued transformations with overlapping window functions. A special class of these transformations are the so-called modulated filter banks, which are characterized by the possibility of an efficient implementation. Among these modulated filter banks, the modified discrete cosine transformation (MDCT), in which the window length L _ _"5 _

can assume values between N and 2N - 1 due to different degrees of overlap.

6 shows the decomposition of a discrete-time input signal x (n) into the spectral components U _k , _m , where m represents the temporal block index, ie the time index after the sampling rate reduction, while k is the frequency index or subband index. The sampling frequencies are the same in all subbands, ie the original sampling frequency is reduced by a factor of N. The filter bank shown in Fig. 6 with filters 60 and downstream downsampling elements 62 provides a uniform band Aufeilung.

In a modulated filter bank, the individual subband filters are created by multiplying a prototype impulse response hp (n) by a subband-specific modulation function, the following rule being used for the MDCT and similar transformations:

The above transformation rule can also deviate from the above equation, e.g. B. if the sine function is used instead of the cosine function, or if "+ N / 2" is used instead of "-N / 2". Use with the alternating MDCT / MDST mentioned later (when using k instead of k + 1/2) is also conceivable.

In the equation above, h _P (n) represents the prototype impulse response. H _k (n) is the filter impulse response for the filter associated with subband k. n is the counting index of the discrete-time input signal x (n), while N is the number of spectral coefficients. The initial values of a real transformation, such as B. the MDCT, which is known to be non-energy-saving, can only be used to a limited extent for applications that require complex Spektra.l components. If, for example, the amounts of the real output values are used as an approximation for the amounts of complex-value spectral components in the corresponding frequency ranges, there are strong fluctuations even with sinusoidal input signals of constant amplitude. Such a procedure accordingly only provides poor approximations for short-term magnitude spectra of the input signal.

In the specialist publication "A Scalable and Progressive Audio Codec", Vinton and Atlas, IEEE ICASSP 2001, May 7-11, 2001, Salt Lake City, an audio encoder with a transformation algorithm is shown, which consists of a basic transformation and a second transformation The input signal is windowed by an Kaiser-Bessel window function to generate successive blocks of sample values, and the blocks of input values are then determined either by means of a modified discrete cosine transform (MDCT) or by means of a modified discrete sine transform (MDST) depending on a shift index This basic transformation process essentially corresponds to the TDAC filter bank described in the cited technical publication by Princen and Bradley, whereupon two temporally adjacent blocks of spectral coefficients are combined into a single complex transformation in such a way that the MDCT block ck represents the real parts of complex spectral coefficients _r while the temporally successive MDST block represents the associated imaginary parts of the complex spectral coefficients. A time-frequency distribution of the amount of the complex spectrum is generated from this, a two-dimensional amount distribution over time being windowed in each frequency band, again with 50% overlapping window functions. Then, by means of the second transformation, an amount matrix calculated. The phase information is not subjected to the second transformation.

The alternating use of the output values of an MDCT as a real and an imaginary part is also introduced in the specialist publication "MDCT Filter Banks with Perfect Reconstruction", Karp and Fliege, Proc. IEEE ISCAS 1995, Seattle, WA, as "MDFT".

It has been found that this approximation of a complex spectrum from a real-value spectral representation of the time-discrete input signal is problematic in that an adequate representation of the amount cannot be obtained for tones of certain frequencies. Thus, the determination of short-term magnitude spectra is only possible to a limited extent even with this transformation.

The object of the present invention is to create an improved concept for generating a complex spectral representation of a discrete-time signal.

This object is achieved by a device for generating a complex spectral representation according to patent claim 1, a method for generating a complex spectral representation according to patent claim 18, a device for coding a discrete-time signal according to patent claim 19, a method for coding a discrete-time signal according to patent claim 20, a device for generating a real spectral representation according to claim 21, a method for generating a real spectral representation according to claim 22 or solved by a computer program according to claim 23.

The present invention is based on the finding that a good approximation for a spectral representation of a discrete-time signal can be determined from a block-wise real-value spectral representation of the discrete-time signal by a first partial spectral coefficient - D -

and / or a second partial spectral coefficient is calculated by combining at least two real spectral coefficients. So that z. B. the real part / or the imaginary part of an approximated complex spectral coefficient for a specific frequency index can be obtained by combining two or more real spectral coefficients, preferably in temporal and / or frequency proximity to the complex spectral coefficient to be calculated. The combination is preferably a linear combination, with the real spectral coefficients to be combined also being in front of the linear combination, ie. H. an addition or subtraction, which can be weighted with constant weighting factors.

At this point it should be pointed out that a linear combination is an addition or subtraction of different linear combination partners, which may or may not be weighted with weighting factors before the linear combination. the weighting factors can be positive or negative real numbers including zero.

In a preferred embodiment of the present invention, the two or more real spectral coefficients, which are combined in order to obtain a complex partial spectral coefficient for a frequency index and a (temporal) block index, are arranged in frequency and / or temporal proximity. The real spectral coefficients with a frequency index that is 1 higher or 1 lower from the current (temporal) block are in frequency proximity. In addition, the corresponding real spectral coefficients from the immediately preceding time block or the immediately following time block with the same frequency index are in close proximity. The real spectral coefficients of the immediately preceding or immediately following time block with a frequency index that is higher or lower by a frequency index than that is also located in temporal and frequency proximity Frequency index of the partial spectral coefficient just calculated.

The combination rule for calculating a partial spectral coefficient preferably varies depending on whether the frequency index is even or odd.

According to the invention, it was found that a combination of real spectral coefficients in temporal and / or frequency proximity to the complex spectral coefficient to be determined is a good approximation of a desired frequency response of the entire arrangement from the device for generating a block-wise real-value spectral representation and the device for Postprocessing the blockwise real value spectral display delivers, the frequency response - which usually has a bandpass character - should have a desired course for positive frequencies, and should be as small as possible or negative for negative frequencies. should be equal to 0. Such a frequency response results from the concept according to the invention and is regarded as advantageous for many applications.

The properties of this frequency response can in preferred embodiments, for. B. can be manipulated by suitable setting of the weighting factors or by appropriate modification of the window functions of the first transformation to generate the real-valued spectral coefficients. The system thus provides many degrees of freedom for adaptation to specific needs, in particular mentioning the possibility of not only combining two real spectral coefficients, but also combining more than two real spectral coefficients in order to achieve an even better approximation to a desired frequency response to achieve the overall arrangement.

Preferred embodiments of the present invention are explained in detail below with reference to the accompanying drawings. Show it : 1 shows a block diagram of the device according to the invention for generating a complex spectral representation;

2a to 2c a representation of the real spectral coefficients adjacent to a partial spectral component for a complex spectral coefficient with frequency index k and block index m;

3 shows a schematic illustration for calculating complex subband signals with a real-valued transformation i and a post-processing transformation T ₂ ;

4 shows a block diagram of the device according to the invention in accordance with a preferred exemplary embodiment of the present invention with critical scanning;

5 shows a block diagram of the device according to the invention according to a further exemplary embodiment of the present invention without critical scanning; and

Fig. 6 shows a known real-valued filter bank with uniform band division.

1 shows a device for generating a complex spectral representation of a discrete-time signal x (n). The discrete-time signal x (n) is fed into a device 10 for generating a block-wise real-value spectral representation of the discrete-time signal, the spectral representation having successive blocks in time, each block having a set of spectral coefficients, as described in more detail with reference to FIGS. 2a to 2b is explained. At the output of the device 10 there is thus a sequence of blocks which follow one another in time of spectral coefficients, which are real value spectral coefficients due to the property of the device 10. This sequence of temporally successive blocks of spectral coefficients is fed into a device 12 for postprocessing in order to obtain a block-wise complex approximated spectral representation which has successive blocks, each block having a set of complex approximated spectral coefficients, a complex approximated spectral coefficient can be represented by a first partial spectral coefficient and a second spectral coefficient, at least the first or the second spectral coefficient being determined by a combination of at least two real spectral coefficients.

FIGS. 2a to 2c together show a sequence of blocks of amounts of real-value spectral coefficients, such as are generated by the device 10 of FIG. 1, m represents a block index, while k represents a frequency index. FIG. 2 shows a block of real-valued spectral coefficients at the time or block index (m-1) plotted along the frequency axis. The block of spectral coefficients comprises spectral coefficients Ui, _m -ι, where i is a running index, while m-1 stands for the block index. In particular, a spectral line with the frequency index i = k and a spectral component with the frequency index i = (k-1) and i = (k + 1) are shown in FIG. 2a.

FIG. 2b shows the same situation, but now for the block m following at the time. Finally, FIG. 2c shows the same situation again, but now for the block index (m + 1). In the sequence of FIGS. 2a, 2b, 2c, this results in a time course which is symbolized by an arrow 20 in FIGS. 2a to 2c.

FIG. 3 shows an alternative representation of the device for generating a complex spectral representation, the discrete-time input signal x (n) in the device 10 for Generating a block-wise real spectral representation, which is designated in FIG. 3 with Ti. It should be pointed out that this is a first conversion of the time signal, which has been windowed in order to be present in blocks, in a spectral representation at the output of the device 10. FIG. 3 shows a snapshot at the time or block index m, that is to say relates to FIG. 2b, which has been described above. The output values of the device 10, that is to say the real-value spectral coefficients, which can be, for example, MDCT coefficients, are fed into the device 12 for post-processing in order to obtain a complex spectrum on the output side, which has a first partial spectral coefficient p _k , _m and one for each frequency index _k comprises second partial spectral coefficients q _k , _m , where p _k , _{m is} the real part and q _k , _{m are} the imaginary part of the complex spectral coefficient for the frequency index k, where m denotes the block index.

According to the invention, real-value transformations in the form of modulated filter banks for the actual spectral decomposition are thus used to generate complex-value spectral components. Real spectral coefficients from temporally successive and / or spectrally adjacent output values of the real-valued transformation, which is denoted by Ti or 10 in FIG. 3, are now used. A real and an imaginary part p, q for a specific frequency index and for a specific (temporal) block index are formed from these, for example. Alternatively, the amount and phase could of course also be generated. Here, special phase relationships of the modulation functions can be used, which are the basis of a modulated filter bank.

In a preferred exemplary embodiment, operation T ₂ or 12, which follows the first transformation, is again an invertible, critically sampled transformation. This results in an overall system, which also has the property of critical scanning and at the same time enables reconstruction from the spectral components obtained.

T ₂ is now a two-dimensional transformation, since in the preferred exemplary embodiment of the present invention, both real-time spectral coefficients that are adjacent in terms of time and that are adjacent in terms of frequency are combined, i. H . since their input values extend along the time and frequency axes, as is shown in FIG. 2a to 2c has been shown. Since a real and an imaginary part arise from each transformation operation using the device 12, a pair of values has to be calculated for a critical sampling only for every second sampling position of the time / frequency level. In a preferred embodiment of the present invention, this is achieved by reducing the sampling rate along the time axis, i. H . Calculation only achieved for every second block of the first transformation Ti. Alternatively, this is done by reducing the sampling rate along the frequency axis, i. H . Calculation only achieved for every second subband i of the first transformation. Again alternatively this is offset, i. H . in the form of a checkerboard pattern, in which every second block and every second band are used alternately.

The transformation coefficients of the second transformation, with which the output values of Ti are each weighted before their summation, that is to say the weighting factors, preferably meet the conditions for the exact reconstruction in accordance with the respective scanning scheme. The system according to the invention contains a number of degrees of freedom which are necessary for optimizing the properties of the overall system, i. H . can be used to optimize the frequency response of the entire system as a complex filter bank.

It should also be noted that critical scanning is not required for some applications is. This can e.g. B. be the case in a postprocessing of the decoded but not yet transformed back into the time domain signals in an audio decoder. In this case, you have a higher degree of freedom in choosing the transformation coefficients in T ₂ . This higher degree of freedom is preferred for a better optimization of the overall behavior.

In the following, using FIG. 4 shows a first exemplary embodiment of the present invention for the detailed regulation of the device 12 for post-processing. It is preferred to distinguish between an even frequency index k and an odd frequency index k + 1. In the case of a straight frequency index, that is to say if P _k , _m and q _k , _{m are} to be calculated (m is the block index and k is the frequency index), the real part p _k , _m is calculated by summing up according to the first exemplary embodiment of the present invention two real-time spectral coefficients successively determined. p _k , _m thus results either from the summation of the spectral coefficient with the index k from FIGS. 2b and 2a or from FIGS. 2c and 2b.

The associated imaginary part q _k , _m is according to the invention either by summing two successive values with the frequency index k-1 either from FIG. 2a, 2b (block m-1 and block m) or the Fig. 2b and 2c (block m and block m + 1) obtained.

For an odd frequency index k + 1, the real part _P k ₊ ι, m is calculated as the difference between two successive values, i.e. as the difference between the spectral coefficients k + 1 of FIG. 2a, 2b or 2b, 2c. The associated I-maginary part q _k + ι, m results from the difference between two successive values with the frequency index k, that is to say as the difference from the real-value spectral coefficients with the index k of FIG. 2a, 2b or 2b, 2c. This results in the transformation function shown in FIG. 4, which is denoted overall by reference numeral 12a, the transformation function having two transformation sub-specifications h _L (m) and h _H (m), which, as shown in FIG is shown in pairs, alternately applied to the output values of the device 10. In particular, the first subfunction h _L (m) has the form {1, 1}, while the second subfunction comprises the form {1, -1}. The notation of the sub-functions h _L (m) and h _H (m) is intended to mean that a sum or difference of the corresponding spectral coefficients is to be formed from two (temporally) adjacent blocks.

The critical sampling is achieved by reducing the sampling rate over time by a factor of 2, as symbolically represented by the device denoted by 12b in FIG. 4. If orthogonality of the second transformation (12a, 12b) is desired, all output values p, q can be standardized by multiplication by the factor 1 / V2.

The second transformation (12a, 12b) which follows the first transformation, which is, for example, an MDCT, extends over the two adjacent bands from which the real part p _k , _m and the imaginary part q _k , _{m are formed} for a frequency index k. In addition, as represented by the functions h _L and h _H , temporally successive real-value spectral coefficients are taken into account in the combination, ie the summation or difference formation.

Since the downstream transformation 12a, 12b in the exemplary embodiment shown in FIG. 4 does not include any degrees of freedom for optimizing the overall system in the sense of adjustable weighting factors contained in the functions h _L and h _H , it is preferred to optimize the overall system using the window function of the first transformation , for example the MDCT, to manipulate, ie in To change compared to a given known window function. This gives a degree of freedom N / 2 with a frequency resolution of N subbands and a window length of L = 2 N values.

In summary, the transformation rule T ₂ shown in FIG. 4 is as follows:

for k straight:

qk, m - U _k -ι, _m + U] ς_ι, _m -l (2)

for k + 1:

P _k + l, m - Ujc + ι _rm - U fc ₊ i, _m -ι (3)

q _k + l, m = u _k , _m - U _k , _m _ι (4)

For canceling the transformation T, as illustrated for FIG. 4 by way of example in the equations (1) to (4), an inverse to the transform rule T ₂ T transformation rule is used ₂ ^-1. If equations (1) to (4) are considered, it can be seen that the real spectral components u _k , _m _ι and u _k , _m from the real part p _k , _m and the imaginary part qk ₊ ι, _m from Equations (1) and (4) can be calculated by solving the two equations (1) and (4) for two unknowns according to the real spectral _coefficients u _k , _m - _ι and u _k , _m sought. Using this inverse combination rule T ₂ ^-1 , knowledge of the sequence of blocks of complex approximated spectral coefficients can be used to trace back to the sequence of real spectral coefficients. can be calculated by performing the inverse combination rule.

An alternative exemplary embodiment is described below with reference to FIG. 5, in which no critical scanning is provided. Here, the output value u _{Itl of} the m th MDCT operation with the frequency _index k is used directly to form the real part. The associated imaginary part is the weighted sum of the MDCT output values surrounding the time-frequency level u _k _ι, _m _ι, u _ι, _m , u _k _ι, _{m +} , u _k , _m -ι, u _k , _{m +} ι, u _{+1 m} _ι, u _{k +} ι, _m and u _{k +} ι, _{m +} ι calculated. A possible combination of the corresponding filters according to FIG. 5 (in the example for k odd) is as follows:

for the real part p:

h _R (m) = {0, 1, 0},

for the imaginary part q:

h _A (m) = {a, -b, a}, h _B (m) = {c, 0, -c}, h _c (m) = {a, b, a}

In the above expression, the values of the coefficients a, b, and c can be used to optimize the overall system, that is, again to achieve a desired frequency response of the overall arrangement, which, as has been explained, is desired, for example, in that for positive frequencies, a bandpass characteristic is present as a frequency response, while the greatest possible attenuation is desired for negative frequencies.

Expressed equation default, represents the transformation rule T ₂ shown in FIG 5, which is of the single filters 50a, 50b, 50c, 50d and an adder 50e loading, as follows: FIG.

for k odd: P _k , _m = u _k , _m ; (5)

9j, _m ^{= a} 'U _k -i, _{m +} ι - b ^" U _k _ι, _m + a ^' U] c-ι, _m _ι +

-c ^• u _k , _{m +} ι c ^• u _k , _m _ι +

a 'u _k _ι, _m -ι + b "u _{k +} , _m + a" u _{k +} ι, _m _ι; (6)

For the calculation of q _k , _m , all real spectral _coefficients adjacent to the real spectral _coefficient u _k , _m in the time-frequency plane are used to a greater or lesser extent by weighting factors a, b, c, as shown in equation (6) ,

It should be noted that the same equations (4) to (6) can be used for an even k. In this case, the weighting factors preferably have the same amounts, but in some cases have different signs.

To reverse the transformation rule shown in FIG. 5, only a trivial operation has to be carried out for the determination of u _k , _m , since this value results directly from equation (5). Since the system shown in FIG. 5 is a non-critically scanned system, the real and imaginary parts are shown redundantly in terms of information. This manifests itself in the inverted transformation rule T ₂ ^_1 in that the real spectral coefficients can be calculated from the real parts alone. Equation (6) therefore does not have to be used for the evaluation. The transformation rule inverse to the transformation rule is thus identical in the embodiment shown in FIG. 5 and given by equation (5).

It should be noted that in the case described above, in which the complex approximated spectral representation is required, for example in a psychoacoustic model, in order to tiser step size, a calculation from the complex approximated spectral representation to the real spectral representation is no longer required. Alternatively, however, there may be cases in which a corresponding inversion is required, in which case the underlying real spectral representation must be calculated from the complex approximated spectral representation.

Depending on the circumstances, the methods according to the invention can be implemented in hardware or in software. The implementation can take place on a digital storage medium, in particular a floppy disk or CD with electronically readable control signals, which cooperate with a programmable computer system in such a way that the corresponding method is carried out. In general, the invention thus also consists in a computer program product with program code stored on a machine-readable carrier for carrying out one or more of the methods according to the invention when the computer program product runs on a computer. In other words, the invention is also a computer program with a program code for performing one or more of the methods when the computer program runs on a computer.

Claims

claims

1. Device for generating a complex spectral representation of a discrete-time signal, with the following features:

means (10) for generating a block-wise real-value spectral representation of the discrete-time signal, the spectral representation having successive blocks in time, each block having a set of real spectral coefficients; and

a device (12) for post-processing the block-wise real-value spectral representation in order to obtain a block-wise complex approximated spectral representation which has successive blocks, each block having a set of complex approximated spectral coefficients, a complex approximated spectral coefficient being represented by a first partial spectral coefficient and a second partial spectral coefficient can be represented, at least one of the first and the second partial spectral coefficient being determined by a combination of at least two real spectral coefficients.

2. Device according to claim 1,

in which the first partial spectral coefficient is a real part of the complex approximated spectral coefficient, and in which the second partial spectral coefficient is an imaginary part of the complex approximated spectral coefficient.

3. Device according to claim 1 or 2,

where the combination is a linear combination. - iy -

4. Device according to one of the preceding claims,

in which the device (12) is designed for post-processing in order to combine a real spectral coefficient of the frequency and a real spectral coefficient of an adjacent higher or lower frequency in order to determine a complex spectral coefficient of a certain frequency.

5. Device according to one of the preceding claims,

in which the device (12) is designed for postprocessing in order to combine a real spectral coefficient in a current block and a real spectral coefficient in a preceding block or a subsequent block to determine a complex spectral coefficient of a certain frequency.

6. Device according to one of the preceding claims, which is designed to work with a critical sampling, such that a real spectral value is generated for each time-discrete sample by the means (10) for generating a block-wise real spectral representation, and that for two real spectral coefficients a complex spectral coefficient is generated.

7. The device according to claim 6,

in which the device (12) is designed for post-processing in order to be active for sampling rate reduction only for every second block of real-valued spectral coefficients, or to be active for sampling rate reduction for every second real spectral coefficient, or in order to alternately for sampling rate reduction only for everyone second block or just for to be active every second real spectral coefficient.

8. Device according to one of the preceding claims,

in which the device (12) is designed for postprocessing in order to sum two real spectral coefficients with the same frequency index from a current block and a block preceding in time for the first partial spectral coefficient with a straight frequency index, and to add two for the second partial spectral coefficient with the even frequency index to sum real spectral coefficients with a frequency index lower by 1 from the α current block and the block preceding in time.

9. Device according to one of the preceding claims,

in which the device (12) is designed for post-processing in order to form a difference between two real spectral coefficients with the odd frequency index from a current block and a block preceding in time for the first partial spectral coefficient with an odd frequency index, and for the second Partial spectral coefficients to form a difference between two real spectral coefficients with a frequency index lower by 1 from the current block and the block preceding in time.

10. Device according to one of the preceding claims,

in which the device (12) is designed for post-processing in order to normalize the first and the second partial spectral coefficients each by a factor 1 / ^ 2.

11. The device according to one of claims 1 to 7, in which the device (12) is designed for postprocessing in order to use a 'real spectral coefficient with the frequency index as the first partial spectral coefficient for a frequency index, and in order to calculate a weighted sum of the real spectral coefficients with adjacent frequency indices of a current block for calculating the second partial spectral coefficient to use one or more preceding blocks or of one or more subsequent blocks, at least two weighting factors not equal to 0.

12. The device according to claim 11,

in which the device (12) is designed for post-processing in order not to use the real spectral coefficient which forms the first partial spectral coefficient for calculating the second partial spectral coefficient.

13. The apparatus of claim 11 or 12,

in which the device is designed for post-processing in order to use the following rule for calculating the second spectral coefficient:

q _k , _m ⁼ 3. 'u _k _ι _{m +} ι - b' u _k _, _m + a 'U _k _ι, _m -ι +

-c "u, _{m +} ι c" U _k , _m -ι +

a 'u _k -ι, _m -ι + b' u _{k +} ι, _m + a "u _{k + im} _ι;

where a, b, c are positive or negative weighting factors, where k-1 is a current frequency index k less 1, where m-1 is a current block index m less 1, where k + 1 is a current frequency index k plus 1, where m + 1 is a current block index m plus 1, and where u _k _ι, _m _ι a real spectrum ral coefficient of a temporally preceding block with a frequency index k-1, where u-ι, _{m is} a real spectral coefficient of a current block with a frequency index k-1, where u _k _ _{1 (m +} ι ^e i ⁿ real spectral _{coefficient of} a temporally following block with a frequency index k-1, where u _k , _ra -ι is a real spectral coefficient with the frequency index k from the block preceding in time, whereby u _{k /} _{m +} ι is a real spectral _coefficient with the frequency index for the block following in time, where u _{k +} ι, _m -i is a real spectral coefficient with the frequency index k + 1 from the previous block, where u _k + ι, _{m is} a real spectral coefficient for the frequency index k + 1 from the current block, and where u _{k +} ι, _{m +} ι is a real spectral coefficient with the frequency index k + 1 from the temporally following block.

14. The apparatus according to claim 13,

in which k signs differ from one or more weighting factors for even and odd frequency indices.

15. The apparatus of claim 13 or 14,

in which the weighting factors are set in order to create a desired frequency response for the device for generating a complex spectral representation.

16. Device according to one of the preceding claims,

in which the device (10) is designed to generate a modified discrete cosine transformation.

17. The apparatus of claim 16, in which the device (10) is designed to generate a modified discrete cosine transformation with a window overlap of 50%.

Method for generating a complex spectral representation of a discrete-time signal, with the following steps:

Generating (10) a block-wise real-value spectral representation of the discrete-time signal, the spectral representation having successive blocks in time, each block having a set of real spectral coefficients; and

Postprocessing (12) the block-wise real-value spectral representation in order to obtain a block-wise complex approximated spectral representation which has successive blocks, each block having a set of complex approximated spectral coefficients, a complex approximated spectral coefficient being representable by a first partial spectral coefficient and a second partial spectral coefficient, wherein at least one of the first and the second partial spectral coefficients is determined by a

Combination of at least two real spectral coefficients are to be determined.

Device for coding a discrete-time signal, with the following features:

a device for generating a block-wise real-value spectral representation of the discrete-time signal, the spectral representation having successive blocks in time, each block having a set of real spectral coefficients; a psychoacoustic module for calculating a psychoacoustic masking threshold depending on the discrete-time signal;

a device for quantizing a block of real-value spectral coefficients using the psychoacoustic masking threshold,

wherein the psychoacoustic module has a device (12) for post-processing the block-wise real spectral representation in order to obtain a block-wise complex approximated spectral representation which has successive blocks, each block having a set of complex approximated spectral coefficients, a complex approximated spectral coefficient by one first partial spectral coefficient and a second partial spectral coefficient can be represented, at least either the first or the second partial spectral coefficient to be determined by a combination of at least two real spectral coefficients.

20. A method for coding a time-discrete signal, comprising the following steps:

Generating a block-wise real-value spectral representation of the discrete-time signal, the spectral representation having blocks which follow one another in time, each block having a set of real spectral coefficients;

Calculating a psychoacoustic masking threshold depending on the discrete-time signal;

Quantizing a block of real-valued spectral coefficients using the psychoacoustic masking threshold, wherein in the computing step, a step (12) of post-processing the block-wise real spectral representation is performed to obtain a block-wise complex approximated spectral representation that has successive blocks, each block having a set of complex approximated spectral coefficients, with a complex approximated spectral coefficient due to a first partial spectral coefficient and a second partial spectral coefficient can be represented, at least either the first or the second partial spectral coefficient being determined by a combination of at least two real spectral coefficients.

21. Device for generating a real spectral representation from a complex approximated spectral representation, the real spectral representation to be determined having blocks which follow one another in time, each block having a set of real spectral coefficients, the complex approximated spectral representation having blocks which follow one another in time, each block has a set of complex approximated spectral coefficients, wherein a complex approximated spectral coefficient can be represented by a first partial spectral coefficient and a second partial spectral coefficient, the complex approximated spectral coefficients being calculated by a transformation rule from the real spectral coefficients, the transformation rule being a combination of at least two real spectral coefficient comprises to at least the first or the second sub-spectral coefficient of a complex approximi to calculate the spectral coefficients with the following characteristic:

a device for carrying out a combination instruction inverse to the transformation instruction (T ₂ ) to calculate the real spectral coefficients from the complex approximated spectral coefficients.

2nd Method for generating a real spectral representation from a complex approximated spectral representation, wherein the real spectral representation to be determined has successive blocks, each block having a set of real spectral coefficients, the complex approximating

Spectral representation has temporally consecutive blocks, each block having a set of complex approximated spectral coefficients, a complex approximated spectral coefficient being represented by a first partial spectral coefficient and a second partial spectral coefficient, the complex approximated spectral coefficients being determined by a transformation rule from the real spectral coefficients have been calculated, the transformation rule comprising a combination of at least two real spectral coefficients in order to calculate at least the first or the second partial spectral coefficients of a complex approximated spectral coefficient, with the following step:

Carrying out a combination rule that is inverse to the transformation rule (T ₂ ) in order to calculate the real spectral coefficients from the complex approximated spectral coefficients.

23. A computer program with a program code for performing the method according to claim 18, claim 20 or claim 22 when the program runs on a computer.