CH708710A1 - Deriving multi-channel signals from two or more base signals. - Google Patents

Abstract

The invention relates to a method for extracting at least one output signal from two input signals. The method comprises the direct extraction of multichannel signals on the basis of correlation comparison which, on the one hand, provides a mathematically exact solution for time-invariant (stationary) signals and exhibits a specific residual behavior for time-variant (non-stationary) signals. This led to the direct detection of a very easy to determine signal for all residuals. This can be used, for example, in audio coding for the efficient reduction of artifacts or color discoloration and other unmasking effects - and leads to the efficient coding of signals of the highest order (such as NHK 22.2).

Description

description

Multichannel signals and in particular three-dimensional signals such as audio signals place high demands on data sets to be transmitted or stored, which it is possible to efficiently reduce.

Commonly known devices or methods for such data reduction are here parametric methods that extract spatial information, for example, using the known from the prior art Fast Fourier Transform (FFT) and then as a permanent data stream, such as together with a mono or stereo signal as downmix signal, transmitted. Such an audio technology is known in particular with MPEG Surround and mathematically represents an adaptive filtering method.

The own unpublished application CH 02 300/12, which is briefly explained by way of example with reference to FIG. 9 proposes, however, for the first time the use of the so-called non-linear inverse coding, this with the additional use of correlation comparison; This approach can, because only the parameters of a spatial model, this does not have to be transmitted permanently, as Fleaderinformation or data pulse, mathematically be considered as a static filtering method.

However, CH 02 300/12 does not provide any explicit technical solution for such a correlation comparison, since devices or methods known from the prior art exist here, for example the Upmix system UPM1 of the British company Soundfield, which, likewise based on FFT , a total of high computational complexity requires. The formation of the output signals takes place here, however, amplitude dependent, which brings the disadvantage of migrating sound sources and temporally shifting artifacts on the individual channels, and overall leads to significant spectral discoloration, which have a disturbing effect in the audio coding.

A simple device or a simple method for such a correlation comparison, which also fulfills the goal of the highest possible psychoacoustic transparency, would therefore be desirable, for example, with regard to the coding of multichannel signals in real time.

If such a simple device or a simple method, as is the case for the present invention, based on a Fourier transform, which is applied to time-varying ("non-stationary") signals, so-called residuals occur. It is desirable in this case to be able to determine these residuals in an encoder in such a general way that not all residuals need to be transmitted, which leads to significant bandwidth savings, for example in audio coding.

The present invention is therefore intended to represent a simple device or a simple method, which makes such a correlation comparison with maximum transparency.

It is not limited to audio signals, although in particular can be optimally applied to such a system. For example, with the subject matter of the invention, video signals can be efficiently compressed and decompressed-or their residuals efficiently minimized.

In general, the subject invention can be applied in the entire signal technology, provided that it is based on a Fourier transform or an inverse Fourier transform, where it leads, for example, to drastic bandwidth savings or allows efficient data extraction.

With the present invention, for example, in the audio encoding, the number of downmix channels can be kept to a minimum, since the mixed channels can in turn be isolated by correlation comparison and thus allow overall efficient storage and transmission of highly complex audio signals.

In particular, such highly complex 3D audio signals, as they are known, for example, with the NHK 22.2 format, can now be combined into corresponding downmix channels, which can be subjected in part or in total to such a correlation comparison.

For example, the middle layer and the top layer of a NHK-22.2 system (or similar format) separately undergo such correlation comparisons, since psychoacoustically horizontal planes are perceived substantially separately, and between these planes, that is, vertically , Diagonals, etc., comes only to a small extent to the formation of phantom sound sources.

Accordingly, it is particularly advantageous to apply such correlation comparisons within horizontal levels. In particular, although the invention is not limited to performing such correlation comparisons on adjacent channels, it is advantageous that strict channel separation is a prerequisite for the clean formation of phantom sound sources in high order signals.

Of course, the invention is not limited to the horizontal, but it can also be used vertical or diagonal or other combinations.

The following documents are to be considered in particular as belonging to the prior art:

EP 1 850 639 describes a static filter which generates a stereo signal from a mono signal. This filter can also be applied to multi-channel signals.

[0017] WO 2009 138 205 describes a static filter which generates a stereo signal from a mono signal. This filter can also be applied to multi-channel signals.

WO 201 1 009 649 describes an extension of the static filters described in EP 1 850 639 and WO 2009 138 205 for adapting the degree of correlation of the respectively generated stereo signal. This extension can also be applied to multi-channel signals.

[0019] WO 201 1 009 650 describes extensions of the devices or methods described in EP 1 850 639 and WO 2009 138 205 and WO 201 1 009 649 in order to optimize the respectively generated stereo signal with regard to the static parameters. These extensions can also be applied to multi-channel signals.

[0020] WO 2012 016 992 describes the first practical use of algebraic invariants generally in signaling technology and more particularly EP 1 850 639, WO 2009 138 205, WO 2011 009 649 and WO 2011 009 650.

WO 2012 032 178 describes the temporal scaling of static filters according to EP 1 850 639, WO 2009 138 205, WO 201 1 009 649, WO 201 1 009 650 and WO 2012 016 992.

Own unpublished application CH 02 300/12, which is briefly illustrated by way of example with reference to FIG. 9, describes extensions of these static filters for their specific application to multichannel signals, this also using direct correlation comparison, for example under direct use of the upmix Systems UPMI of the British company Soundfield can take place.

FIAMASAKI KIMIO ET AL :. "The 22.2 Multichannel Sound System and Its Applica- tion", AES CONVENTION 1 18, MAY 2005, describes a channel-based playback format of highest order and spatial resolution.

MPEG Surround circumscribes as a standard the use of so-called parametric method for the transmission of multi-channel signals based on a mono or stereo signal.

Disclosure of the invention

In its own unpublished application CH 02 300/12, see Fig. 9 and below, an extraction of multi-channel signals is proposed based on correlation comparison; however, the document does not provide an explicit technical solution for such a correlation comparison, since devices or methods known in the art exist.

Such a correlation comparison makes, for example, the upmix system UPM1 the British company Soundfield, which, also based on the Fast Fourier Transform (FFT, see below), requires a total of high computational effort. The formation of the output signals takes place here, however, amplitude dependent, which brings the disadvantage of migrating sound sources and temporally shifting artifacts on the individual channels, and overall leads to significant spectral discoloration, which have a disturbing effect in the audio coding.

Such a correlation comparison can for example be applied to a so-called downmix, in which the same signal or similar signal components have been added to further original or successively formed channels, wherein one or more levels of one or more original or successively formed signals can be known.

In the following, as part of the subject invention, such a correlation comparison of two signals L, <'> and R, <'>, in which identical identical signal components x (t) and y (t), which for the short-term cross-correlation

5T f <x (t) ymt> '"rym, rr τ have the degree of correlation +1, on the one hand represents a mathematically exact solution for time-invariant (stationary) signals, and has a specific residual behavior in time variants (non-stationary) signals (where Residual represents the difference between the original, non-stationary signal section and its Fourier transform).

The possible recovery of the corresponding residual is also shown as part of the subject invention.

Furthermore, as part of the subject of the invention, which makes use of the specific residual behavior, an approximate extraction method for residuals is proposed, as long as the residuals of the overall system are known.

Consider two channels L, <'>, R, <'>, I <i <n, which have similar signal components C *, where:

3 Li + Ci li '(t) li <*> (t) + Ci <*> (t)

Ri

Ri + Ci

= ri

'(t) ri (t) + Ci (t)

For the time-dependent signals I, <'> (t) and r {(t), the Fourier series are now determined in each case. Accordingly, for the synthesis, k = ..., - 1,0, 1, ... oo li (t) = ^ xke <lkü> ot> k = - °°

· / M -

/ £ Σ = - «

Vk e ikü) 0t and for the analysis

[0033]

(t) e ~ <ikc> ° o * dt

<Γ> ° / 2

(t) e-ifco) 0tdt and in practice for the discrete Fourier transforms (DFT), from which the Fast Fourier Transforms (FFT) can be directly derived, where now k = 0, ..., N-1:

N-1

L'i (fc) = e <~ l ~> ü <mk> m-0 N-1

R- (k) = rl m) e ~ <l ~> N <mk> m = 0

The real parts of U R, and C, can then be recovered for stationary signals for all k = 0, N-1 according to the following rules:

1. Determine the signs of the real parts of L, <'> (k) and R, <'> (k).

2. If the signs are identical for k, determine

- the magnitudes of the real parts of L, <'> (k) and R, <'> (k),

- the minima or maxima of these amounts of the real parts of L, <'> (k) and Rf (k).

- Choose in each case as real part for C, (k) the real part of L, <'> (k) or R, <'> (k) on which this minimum is based.

- Subtract the real part of C, (k) the real part of L, <'> (k) or R, <'> (k) underlying the maximum and, if the real part of L, <'> (k), choose this maximum otherwise the result of this subtraction as a real part for L, (k), otherwise, if the real part of R, <'> (k) underlies this maximum, the result of this subtraction as a real part for R, (k).

- Set the not yet determined real part of L, (k) or R, (k) equal to zero.

4 3. If the signs of the real parts of L <'> (k) and R, <'> (k) are not identical, set C, (k) equal to zero and set L (k) = L · <'> (k ) and Rj (k) = R, <'> (k).

The imaginary parts of L ,, R, and C, can be recovered for stationary signals for all k = 0, ..., N - 1 according to the following rules:

1. Determine the signs of the imaginary parts of L, <'> (k) and Rf (k).

2. If the signs are identical for k, determine

the magnitudes of the imaginary parts of L, k (k) and Rf (k),

- the minima or maxima of these amounts of imaginary parts of L, <'> (k) and Rf (k).

- Choose as imaginary part for C, (k) the imaginary part of L, <'> (k) or R, <'> (k) on which this minimum is based.

Subtract the imaginary part of Q (k) from the maximum underlying imaginary part of L, <'> (k) or R, <'>

(k) and, if the imaginary part of L, <'> (k) underlies this maximum, choose the result of this subtraction as an imaginary part for L, (k), otherwise, if the imaginary part of R <'> (k) Maximum, the result of this subtraction as an imaginary part for R, (k).

- Set the not yet determined imaginary part of L, (k) or R, (k) equal to zero.

3. If the signs of the imaginary parts of L, <'> (k) and R, <'> (k) are not identical, set C, (k) equal to zero and set L, (k)

= U <'> (k) and Rj (k) = R, <'> (k).

Finally, in order to obtain L, R, and C, for synthesis for the time-dependent signals, k = ..., -1,

0 1 k (t) = ^ fke <ik> "o * fe = -oo <r> << (t)> = Σ gke <iko}> ^ k = - °°

(t) = hke <ikü> ot>

(or in practice for analysis by discrete Fourier transforms (DFT), k = 0, ..., N-1,

N -l

Σ. , h (m) e <1> m = 0

N-1 i,

Ri (k) = rt (m) e. , 2π <1> N mk i, -2π j-mk,

Σ .2π, c £ (m) e <l> N <mk> m = 0 N -lm = 0 from which the Fast Fourier Transforms (FFT) can be directly derived) for the synthesis the coefficients fk, gk, hk are determined, k = 0, 1, ..., according to the analysis

5 <> <Γ> ° / 2ί = ί h e- ^ dt <ι> °

< "Γη> / 2

'0 / / 2

9k = ί Ά (t) e ~ <ik <> ° <ot> dt <ι> ° h

'

J (t) e <~ ikü> 0> 'dt

•

// 2 or for the synthesis according to the inverse discrete Fourier transformation (IDFT), from which the Inverse Fast Fourier Transforms (IFFT) can be derived directly, k = 0, N-1,

N m- 0 N-1 m = 0

"JV-1

1, • 22n7 ci (m = ij <c> A; i <k> ^ <e N> ,. • 22n7Γ A; n / c m = 0

Moreover, since a number of audio codecs for lossless or lossy compression of audio signals already make use of the Fourier transformation or FFT, the rules described above for obtaining the real parts or imaginary parts can be integrated directly into such audio codecs with little computation effort , or signals derived from such audio codecs that can be subjected to these rules for obtaining the real parts or imaginary parts.

The schematic sequence of such a correlation comparison is shown by way of example in FIG. 15:

For each one channel of the time-dependent downmix signal L, (t), R, (t), a Fast Fourier Transform (FFT) is executed first, and thus the frequency-dependent complex-valued signal descriptions L, (k) and R, ( k). Now the rules for obtaining the real parts and the imaginary parts of L ,, R, and C are applied to them. Finally, an Inverse Fast Fourier Transform (IFFT) is applied to the resulting signal descriptions L, (k), R, (k) and C, (k). The result is the time-dependent signals C, (t), l, (t) and ri (t).

For non-stationary signals occurs in this form of the correlation comparison, a residual Δ, which generally has the following behavior:

Li = L * + ΔR, = Ri * + ΔCi = C * - 2Δ

This residual plays psychoacoustically, as far as it concerns the pure reproduction of Lj, Rj and C, according to an inverse discrete Fourier transform (IDFT), from which the Inverse Fast Fourier Transform (IFFT) can be directly derived normative interception situation (within the sweet spot) does not matter because the residual is wiped out. Outside of the "sweet spot", as is the case with daily life interception situations and with non-normative loudspeaker constellations, however, clearly audible artifacts can occur which must be avoided.

In particular, in the spatial coding of multi-channel signals, provided that these signals thus extracted, especially outside the normative listening situation (outside of the "sweet spot"), discoloration of the tone and other unmasking effects result.

Depending on the application, it is thus desirable to determine such residuals, for example in an encoder, and, if appropriate, in order to reduce the number of transmission channels as a whole, if possible

6 to approximate to discoloration of the timbre and other unmasking effects in the further spatial coding to minimize or eliminate in terms of subjective perception.

The residual Δ itself can be approximately frequency-dependent (the Fourier transformation for L, or R, or C, is already known, and it is thus described only in the same manner as for L, or R, or C. to carry out the Fourier transformation for Li * or R * or C *) for each frequency k as follows (in the case of frequency-dependent computation, an inverse discrete Fourier transformation (IDFT), if any, for A (k)) may be used immediately to derive the Inverse Fast Fourier Transform (IFFT), as described for L, or R, or C, to execute):

A (k) = Li (k) - Li * (k) or

A (k) = R i (k) - R * (k) or

A (k) = 1/2 * (C * i (k) -Ci (k))

This means that to obtain a residual-free signal by means of correlation comparison from two channels L, R 'and R 1, 1 <i <n, which have similar signal components C *, the associated residual Δ must also be known which is determined approximately within the encoder, which represents a major limitation in audio coding, for example, since the absolute number of channels to be transmitted to the decoder can not be reduced to the effect that each such correlation comparison must also be accompanied by such a residual. Alternatively, the residuals can also be determined as a function of time if the common signal C, (t) determined by correlation comparison or the first individual signal L, (t) determined by correlation comparison or the second individual signal R, (t) determined by correlation comparison is time-dependent.

In principle, such non-residue signals can be obtained by simple subtraction or addition both frequency-dependent and time-dependent:

Li * = Li - ΔRi * = Ri - ΔC * = Ci + 2Δ

However, it can be shown that, for example, for adjacent channels L 1, CM, R 1, C 2 and B, where G <'> = L * + CN *

Rn = Ri * + Ch * + Ci2 *

R | 2 <=> Ri * + Cil * + Cj2 *

Bi <'> = B * + Ci2 * does not derive from a residual Δ-ι, which results from the correlation comparison between L, <'> and Rh <'>, in a linear form the residual Δ2, which results from the correlation comparison between Ri2 < '> and B, <'> results, can be deduced.

An ideal approximate determination, which also represents the drastic reduction in the number of residuals to be transmitted, consists in the following consideration:

If n Residuale Δ-ι, Δ2, Δ3, Δ4, ..., Δη determined, and apply to the differences

Δ2- Δι = η3- ηη

Δ3 - Δ2 = η4- ηι

Δ4- Δ3 = η5 ~ Τ | 2

Δη Δ - Τ] ι ~ Τ] η-2

Δι - Δ "= η2- ηη-ι then the relationships can be

7 Δχ = η2-ηη-χ + Δη

Δ2- Δχ = Δ2- (η2-ηη-ι + Δη) = η3- ηηbzw.

Δ2 = η3-ηη + (η2-ηη-ι + Δ ")

Δ3-Δ2 = Δ3- (η3-ηη + (η2-ηη-ι + Δη)) = η4- ηι or

Δ3 = η4-ηι + (η 3 -ηη + (η 2 -ηη-ι + Δη))

Δ - Δ3 == Δ4- (η4-ηι + (η3 ~ ηη + (η2-ηη-ι + Δη))) = η5-η2bzw.

Δ4 = η5-η2 + (η4-ηι + (η3-ηη + (η2-ηη-ι + Δη)))

Δι - Δη = (η2- ηη-ι + Δη) - Δ "= η2- ηη-ι or.

Δι = (η2- ηη-ι + Δη) derived.

Thus, for example, (η 2 - ηη.ι + Δη) is a term that is included in all Residuals. The same reasoning applies to every Δ ,, i = 1, ..., n. Since n remains small in practice, it follows that with every term so determined, each residual can be approximated with high accuracies.

Now set

8 Δ2- Δι = η3- ηη = ax

Δ3-Δ2 = η4-ηι = a2

Δ4 - Δ3 = η5 - η2 = a3

Δ5-Δ4 = η6-η3 = a4

Δβ - Δ5 = Τ | 7 "Τ) 4 <=> 3-5

Δ7 - Δδ = ηβ - Ή5 = a6

Δ3- Δ7 = η9 ~ η e <=> 3.7

Δ9-Δ3 ~ ηιο -η7 <=> ag

Διο - Δ 9 = ηιι - η3 = a9

Δι - Δη = η2-ηη-ι = gives η3 = Πη + ai ηε = Ήη + ax + a4η9 <+> ax + a4 + a7bzw. η4 = ηι + a2η7 = ηι + a2 + as ηιο = ηι + a2 + as + a3bzw. η5 = η2 + a3Όβ = η2 + a3 + a6 ηιι = η2 + a3 + a3 + a9

From the relationships Tin = Ήη + ai + a4 + a7 + ■■■ + Sn-: Tjn-i = Tjn-i + an + a3 + a6 + ... + 1

Tjn-2 = T) n-2 + 3n-1 + 3-2 + 85 + ... +

2bzw. ln-3 bZW. an-4

9, the "difference rule for residuals" can now be

(a-1 + 84 + 87 + ... + an-2) + (an + 83 + 36 + ... + an-3) + (an-i + 82 + 85 + ... + an-4) = 0, which means that a-ι, a2, a3, ..., anpsychoacoustically within a normative listening situation (within the sweet spot) have an ideal residual behavior in that they cancel each other out.

From the above "difference rule" can be derived directly the following "addition theorem for residuals", namely Δ1 + Δ2 + Δ3 + ... + Δη = n * (η2- ηη_ι + Δη) which means that the mean of all Residuale at the same time This is included and can also be easily calculated, for example, within an encoder.

If now the residual correction instead of the residual Δ-ι, Δ2, Δ3, ..., Δηnunmehr made on the basis of their mean, not only discoloration of the timbre and other unmasking effects can be specifically minimize, but at the same time a drastically reduced number of Channels, for example, from an audio encoder to an audio decoder.

2, whose circumscribed triangle with its vertices denotes the number of downmix channels and whose inscribed, dashed triangle the number of additionally extracted channels by means of correlation comparison (which are subsequently subtracted from the associated downmix channels). to approximate all the original channels of the circumscribed triangle), when transmitting the average of all residuals, extract at most three channels of a maximum of six channels having significantly less artifacts or discoloration of the timbre and other unmasking effects. In other words, the vertices of the circumscribed triangle describe the three downmix channels of a six channel multi-channel signal. A vertex of the inscribed triangle describes a, the two adjacent downmix channels mixed channel of the multi-channel signal. Such a further channel lying between two downmix channels can be obtained by a correlation comparison between the two adjacent downmix channels (the vertices of the circumscribed triangle) by extracting this signal contained in both downmix channels, as well as in each case the sum of the two original, external corner signals the downmix with its adjacent signal, which lies in the middle of the nearest corner of the triangle considered to be the respective corner signal (not on the side on which the first correlation comparison was performed!). If a correlation comparison is now also carried out for the two downmix channels of this newly viewed page, the signal contained in both downmix channels is again extracted. This can be subtracted from the nearest sum of the first correlation comparison, and then gives the original corner signal. If this is done for all three adjacent pairs of downmix channels, one obtains again the six channels of the multichannel signal. Since, in addition to the downmix signals, if not stationary, three residuals would have to be transmitted in addition to the three downmix channels for exact calculation of the six multichannel signals, the amount of data to be transmitted would again be equal to the transmission of the multichannel signal. Therefore, the average of all three residuals is now transmitted and the six channels of the multi-channel signal obtained from the correlation comparison are corrected on the basis of this averaged residual.

3, whose circumscribed square with its vertices denotes the number of downmix channels, and whose inscribed, dashed square the number of additionally extracted by means of correlation comparison channels (which are then subtracted from the associated Downmixkanälen to to approximate all the original channels of the circumscribed square), to extract a maximum of eight channels transmitting substantially all artifacts from at least four downmix channels, which have significantly lower artifacts or discoloration of the timbre and other unmasking effects.

4, whose circumscribed pentagon with its vertices designates the number of downmix channels and whose inscribed, dashed pentagon the number of additionally extracted by means of correlation comparison channels (which are then subtracted from the associated downmix channels to all to obtain approximate original channels of the circumscribed pentagon), when transmitting the average of all residuals, to extract from at least five channels a maximum of ten channels that have significantly lower artifacts or discoloration of the timbre and other unmasking effects.

2 to 4 have purely combinatorial meaning and are not to be confused with specific speaker positions.

This scheme can be extended to infinity, but the calculated mean of all residuals due to the above considerations increasingly leads to artifacts or discoloration of the timbre and other unmasking effects.

Further channels can be calculated in all cases with a linear or non-linear inverse coding of existing or successively derived channels or other known from the prior art further spatial coding method approximatively:

In the following, "inverse coding" is understood to mean a technical sequence which is one or more methods or one or more devices of the claims of the applications EP 1 850 629 or WO 2009 138205 or WO 201 1 009 649 or WO 2011 009 650 or WO 2012 016 992 or WO 2012 032 178, the above-mentioned documents being hereby incorporated by reference. In particular, the inverse coding is in these

10 documents. While in the linear inverse coding the two output channels appear uniformly amplified, Fig. 9 shows the example of a non-linear inverse coding characterized in that the two output channels are not uniformly amplified.

The Downmixkanäle or Residuale, which should serve the most efficient storage and transmission of audio data, such as between an encoder and a decoder, can be with a known from the prior art corresponding lossless or lossy base audio codec (examples of a Opus or the MPEG standards MP3, AAC, HE-AAC, HE-AAC v2 and USAC) additionally compress and decompress such Base Audio Codecs, whereby the Base Audio Codec used in each case can additionally be optimized with regard to the overall underlying coding method (" Tuning ").

In particular, since a number of audio codecs for lossless or lossy compression of audio signals already use the Fourier transform or Fast Fourier Transform (FFT), the rules described above for obtaining the real parts or imaginary parts of Signals are integrated directly into these audio codecs, or signals are derived from such audio codecs that can be subjected to these rules to obtain the real parts or imaginary parts. The total amount of computation required can thus be significantly reduced.

DESCRIPTION OF THE DRAWINGS

Various embodiments of the present invention will now be described by way of example, with reference to the following drawings:

Fig. 1 shows the eight possible cases of an unexpectedly handy algorithm for correlation comparison of two signals based on a Fourier transform, wherein in each case those signal components which have the degree of correlation +1 for the short-term cross-correlation are exactly extracted for stationary signals. For non-stationary signals, a residual correction is also possible according to the disclosure of the invention exactly or also based on the mean value of all residuals.

Figure 2 geometrically illustrates the combinatorial application of such a correlation comparison to a corresponding three-channel downmix represented by the circumscribed triangle.

FIG. 3 geometrically illustrates the combinatorial application of such a correlation comparison to a corresponding four-channel downmix represented by the circumscribed square.

FIG. 4 geometrically illustrates the combinatorial application of such a correlation comparison to a corresponding five-channel downmix represented by the circumscribed pentagon.

Fig. 5 shows a 5.1 -Surround arrangement according to ITU-R BS.775-1.

FIG. 6 shows an inventive coding of multi-channel signals on the basis of correlation comparison and possible residual correction or additional spatial coding.

Fig. 7 shows a NHK-22.2 arrangement.

FIG. 8 shows the application of FIG. 6 to an NHK-22.2 middle layer signal with simultaneous use of two inverse codings for FL, FLc and FR, FRc.

Fig. 9 shows the example of a non-linear inverse coding according to the unpublished application CH 02 300/12.

Fig. 10 shows the application of Fig. 6 to a NHK-22.2 top-layer signal with simultaneous correlation comparison to obtain the TpC. This correlation comparison takes place in the psychoacoustically uncritical frontal axis above the head, in which an exact localization remains difficult.

FIG. 11 shows the application of FIG. 6 to a NHK-22.2 top-layer signal with simultaneous panning of the sum of TpC and TpFC. This panning takes place in the psychoacoustically uncritical frontal axis above the head, in which an exact localization remains difficult. Alternatively or additionally to the panning, a linear or non-linear inverse coding can also be carried out.

FIG. 12 shows an encoder module based on the subject invention, which calculates both a downmix and the residual of the associated correlation comparison according to the rules for acquisition

11 of the real and imaginary parts of signals, see disclosure of the invention. All output signals were previously Fourier Transform (FFT).

Fig. 13 shows an encoder for an NFIK 22.2 top layer signal based on the subject invention, which calculates both the total downmix and the average of all residuals calculated by the encoder chips.

Fig. 14 shows a decoder based on the total downmix and the transmitted average of all calculated by the encoder components Residuale by correlation comparison, according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, and by means of sum and difference the original Input signals of the encoder calculated approximately. Finally, the Inverse Fast Fourier Transforms (IFFT) are determined.

Fig. 15 shows the principle of the illustrated correlation comparison, according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention. Prior to this correlation comparison, a Fast Fourier Transform (FFT) is performed, followed by an Inverse Fast Fourier Transform (IFFT).

DETAILED DESCRIPTION

Application of the subject of the invention to a 5.1 surround signal:

A first, simple example of an application of the subject invention to a 5.1 surround signal in accordance with ITU-R BS.775-1, see FIG. 5, with additional use of an inverse coding, represents for the channels L *, R *, C *, LS *, RS the summation (the «downmix»)

L <'> = (L * + 1Λ / 2 * LS *) + 1/2 * C *

R <=> (R * + 1Λ / 2 * RS *) + 1/2 * C *.

L <'> and R <'> can be tuned by means of a base audio codec, which can be specially adapted for this purpose, for efficient storage or transmission, see FIG. 6 (in the present example still an additional spatial encoding and decoding based on the so-called linear or non-linear inverse coding takes place), compress and then decompress.

In the encoder, the left signals LS * and L * are first combined to form a common left signal (L * + 1 / V2 * LS *), and the right signals RS * and R * are combined to form a common right signal (R *). + 1 / V2 * RS *). To determine the parameters that psychoacoustically separate the signals (LS * and L * or RS * and R *) from the common signal ((L * + 1 / V2 * LS *), (R * + 1 / V2 * RS *)), an inverse coding method is used, as disclosed in one of the patent applications WO 2009 138 205, WO 201 1 009 649, WO 201 1 009 650, WO 2012 016 992 or WO 2012 032 178. The disclosure of these applications is for the determination of the parameters which are used for the psychoacoustic separation of the signals (LS * and L * or RS * and R *) from the common signal ((L * + 1 / V2 * LS *), (R * + 1 / V2 * RS *)) are necessary, inserted here. By correlation comparison according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention of L <'> and R <'> can now be a signal 1/2 * C (which is then multiplied by a factor of 2) as also two signals L and R are extracted. In the encoder, an estimate of the signals (L * + 1 / V2 * LS *) or (R * + 1 / V2 * RS *) or C * is determined via the method according to the invention of the correlation comparison between L <'> and R <'> , and the difference with the actual signal is formed to determine the residual Δ. The encoder now outputs L <'>, R <'>, Δ and the parameters for the separation from the common left signal to the two left signals, and the parameters for the separation from the common right signal to the two right signals.

The correlation comparison can make use of a Fourier transformation already carried out in the base audio coder, whereby the overall computational effort required can be significantly reduced.

In the decoder, on the basis of the correlation comparison of L <'> and R <'> according to the invention, an estimate of the signals (L * + 1 / V2 * LS *), (R * + 1 / V2 * RS *) and C * determined. On the basis of the optionally transmitted residual Δ for this correlation comparison, for example from an encoder to a decoder, the original signals C *, (L * + 1 / V2 * LS *) and (R * + 1 / V2 * RS *) can now be used. from whose estimates C, (L + 1 / V2 * LS) and (R + 1 / V2 * RS) are reconstructed according to the following formulas:

12 C <*> = C + 2Δ

(L

+ 1 / J <~> 2 * LS <*>) =

(L + l / 2 *

LS

Δ

(R <*> + 1 / Γ2 * RS

(R + l / v <r> 2 * RS) - Δ

However, due to the small number of channels such a residual determination would lead to no real compression, but rather serves to illustrate the basic residual behavior that goes along with such a correlation comparison and can be corrected accordingly. In this case, the decoded common left signal (L * + 1 / V2 * LS *) corresponds to the original channel combination (L * + 1 / V2 * LS *) of the multichannel signal. If in a multi-channel signal with more channels an additional residual-based averaged residual is used for the correction, the decoded common left signal (L * + 1 / V2 * LS *) is just an estimate (analogous to the common right signal).

On the basis of the parameters transmitted for the separation, approximately two left channels L * and LS * are now calculated approximately for the common left signal (L * + 1 / V2 * LS *) obtained by correlation correction and possibly by residual correction, and for the common right signal (R * + 1 / V2 * RS *) obtained by correlation comparison and possibly by residual correction, two right channels R * and RS * are now calculated. This can be done by means of a linear or non-linear coding, as e.g. is shown in Fig. 9 done. The parameters received by the encoder are φ (angle between sound source and main microphone axis), α (certain left aperture angle), β (certain right aperture angle), f (directional characteristic of the mono signal to be stereophoned), γ (amplifier for changing the degree of correlation or damping to change the degree of correlation) or p (attenuation to change the degree of correlation) and s (time parameters) (or parameters derived from these parameters) are used in the decoder to obtain psychoacoustically optimum delays and gains of the input signal and thus an input signal into two adjacent ones Split channels.

Application of the Subject of the Invention to an NHK-22.2 Middle Layer Signal (See Figures 7, 8 and 15):

A second, complex example of an application of the subject invention, here according to FIG. 8, to a NHK22.2 middle-layer signal as a multi-channel signal, see FIG. 7, for the channels FL, FR, FC, BL, BR, FLc, FRc, BC, SiL and SiR the summation (the channels of the downmix signal)

FL '= FL + 1 / V 2 * FLc + 0.5 * FC + 0.5 * SiL

FR '= FR + 1Λ / <"> 2 * FRc + 0.5 * FC + 0.5 * SiR

BL '= BL + 0.5 * SiL + 0.5 * BC

BR '= BR + 0.5 * SiR + 0.5 * BC where FL <'>, FR <'>, BL <'>, BR <'> correspond to the vertices of the circumscribed ouadrate of Fig. 3 in Fig. 8.

The channels FL and FLc are merged in the encoder before the correlation comparison made to calculate the residual, to a common front left channel, and the parameters necessary for the separation are determined. The channels RL and RLc are merged in the encoder before the, to calculate the Residuale performed correlation comparison to a common front right channel, and determines the parameters necessary for the separation. This is e.g. as described in connection with the assembly of the channels LS * and L * in the 5.1 system. Accordingly, in the decoder, based on the channels of the downmix signal FL <'>, FR <'>, BL <'>, BR <'> with the correlation comparison and possibly a correction by the mean residual Δ, the channels (FL + 1 / V2 * FLc), (FR + 1 / V2 * FRc), FC, BL, BR, BC, SiL and SiR of the multi-channel signal. Thereafter, analogous to the channels L *, LS *, R *, RS * of the 5.1 system, the channels FL, FR, FLc, FRc from the channels (FL + 1 / V2 * FLc), (FR + 1 / V2 * FRc) determined. FL <'>, FR <'>, BL <'>, BR <'> and, if appropriate, the mean value Δ of all residuals Δ-1, Δ2, Δ3, Δ * can be determined using a base audio codec, which is specific for this purpose can be adapted ("Tuning"), for the purpose of efficient storage or transmission, for example between an encoder and decoder, see Fig. 6 (in the present example, an additional spatial encoding and decoding using the so-called

13 non-linear inverse coding between the channels FL and FLc or between the channels RL and RLc takes place), compress and then decompress.

Likewise, the system described below can make use of a Fourier transformation already carried out in the base audio coder, whereby the overall computation effort required can be considerably reduced.

By correlation comparison of FL <'> and FR <'> according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, now a signal 0.5 * (FC - 2 * Δ-ι) are extracted ( which is then multiplied by a factor of 2) as well as two signals (FL + 1 / V2 * FLc + 0.5 * SiL) + Δ-1 and (FR + 1 / V2 * FRc + 0.5 * SiR) + Δ · |. See Fig. 8.

By correlation comparison of FR <'> and BR <'> according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, now a signal 0.5 * (SiR - 2 * Δ2) are extracted (which multiplied by 2) as well as two signals (FR + 1 / V2 * FRc + 0.5 * FC) + Δ2 and (BR + 0.5 * BC) + Δ2. See Fig. 8.

By correlation comparison of BR <'> and BL <'> according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, now a signal 0.5 * (BC - 2 * Δ3) are extracted (which multiplied by 2) as well as two signals (BR + 0.5 * SiR) + Δ3 and (BL + 0.5 * SiL) + Δ3. See Fig. 8.

By correlation comparison of FL <'> and BL <'> according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, now a signal 0.5 * (SiL - 2 * Δ4) are extracted (which multiplied by 2) as well as two signals (FL + 1 / V2 * FLc + 0.5 * FC) + Δ4 and (BL + 0.5 * BC) + Δ4.

See Fig. 8.

With the thus extracted signals 0.5 * (FC - 2 * Δ-1), 0.5 * (SiR - 2 * Δ2), 0.5 * (BC - 2 * Δ3), 0.5 * (SiL - 2 * Δ4) if the residuals Δ-1, Δ2, Δ3, Δ4 are not known, then now approximately calculate all other signals FL + 1 / V2 * FLc, FR + 1 / V2 * FRc, BR, BL:

FL + 1 / V2 * FLc s (FL + 1 / V2 * FLc + 0.5 * SiL) + ^ - 0.5 * (SiL - 2 * A4) = (FL + 1 / V2 * FLc + 0.5 * FC) + Δ4- 0.5 *

(FC - 2 * Δ-1)

FR + 1 / V2 * FRc s (FR + 1 / V2 * FRc + 0.5 * SiR) + ^ - 0.5 * (SiR - 2 * AZ) = (FR + 1 / V2 * FRc + 0.5 * FC) + Δ2- 0.5 *

(FC - 2 * Ai)

BR s (BR + 0.5 * BC) + Δ 2 - 0.5 * (BC - 2 * Δ 3) = (BR + 0.5 * SiR) + Δ 3 - 0.5 * (SiR - 2 * Δ 2)

BL s (BL + 0.5 * BC) + Δ 4 - 0.5 * (BC - 2 * Δ 3) = (BL + 0.5 * SiL) + Δ 3 - 0.5 * (SiL - 2 * Δ 4)

From the double solution paths it follows that the correlation comparison does not necessarily have to be made for all three possible output signals, see also FIG. 14, but may also contain fewer output signals. This results in a myriad of different possible combinations, which can be easily derived from the above equations.

By the way, the same remarks also apply to systems with residual corrections.

If the residuals Δ-1, Δ2, Δ3, Δ4 are known, the following boldly marked residual corrections result (although in such a system no compression can be achieved, since at least one such residual must ultimately be assigned to each correlation comparison):

FL + 1 / V2 * FLc = (FL + 1 / V2 * FLc + 0.5 * SiL) + ^ - 0.5 * (SiL - 2 * Δ4) - ^ - Δ4 = (FL + 1Λ / 2 * FLc + 0.5 * FC ) + Δ4

- 0.5 * (FC - 2 * Δ-ι) - Δ4- ΔΊ

FR + 1 / V2 * FRc = (FR + 1 / V2 * FRc + 0.5 * SiR) + ^ - 0.5 * (SiR - 2 * Δ2) - ^ - Δ2 = (FL + 1 / V2 * FLc + 0.5 * FC ) + Δ2- 0.5 * (FC - 2 * Δ ^ - Δ2-Δ-ι

BR = (BR + 0.5 * BC) + Δ2- 0.5 * (BC - 2 * Δ3) - Δ2 - Δ3 = (BR + 0.5 * SiR) + Δ3 - 0.5 * (SiR - 2 * Δ2) - Δ3 - Δ2BL = (BL + 0.5 * BC) + Δ4- 0.5 * (BC - 2 * Δ3) - Δ4 - Δ3 = (BL + 0.5 * SiL) + Δ3 - 0.5 * (SiL - 2 * Δ4) - Δ3 - Δ4

If now the residual corrections are not made on the basis of the residuals Δ4, Δ2, Δ3, Δ4, but on the basis of the mean value A of all residuals, the residual corrections denoted bold are now represented by the expression

- replace 2Δ. Compared to signals without residual correction, this leads to markedly reduced artefacts or discoloration of the timbre and other unmasking effects, without having to transfer all residual Δ-1, Δ2, Δ3, Δ4, for example from an encoder to a decoder. This results in a drastic reduction of the bandwidth.

If other spatial encodings and decoders are to be used, such as the so-called non-linear inverse encoding according to our own unpublished application CH 02 300/12, see FIG. 9, these can be directly integrated into the above considerations according to FIG.

For example, FL and FLc or FR and FRc can be advantageously obtained by such a non-linear inverse coding of the absolute or approximately obtained signals for (FL + 1 / V2 * FLc) or for (FR + 1 / V2 * FRc) also approximately gain:

For example, the left output signal for FL of an arrangement according to FIG. 9 is amplified by the factor 1 (60001), but the right output signal for FLc of such an arrangement is multiplied by the factor 1 / V2 (60002). Similarly, the right output signal for FR of an arrangement according to FIG. 9 is amplified by the factor 1 (60002), whereas the left output signal for FRc of such an arrangement is multiplied by the factor 1 / V2 (60001).

An NHK-22.2 middle-layer signal can thus be significantly reduced in terms of data to be stored or transmitted, for example between an encoder and a decoder, in the sense of FIG.

Application of the subject invention to an NHK 22.2 top layer signal without TpC (see FIGS. 7 and 10 to 15):

The principle of action just described for a NHK-22.2 middle-layer signal can be easily transferred to a NHK22.2 top-layer signal, provided the following equations are made in the above example:

TpFL = FL + 1 / V <~> 2 * FLc TpFR = FR + 1 / V <~> 2 * FRc TpFC = FC TpSiR = SiR TpBR = BR TpBC = BC TpBL = BL TpSiL = SiL

An additional spatial encoding and decoding for TpFL and TpFR is therefore eliminated.

However, in such a transmission the TpC is neglected, which plays an essential role, for example, in NHK-22.2 top-layer signals.

Application of the subject invention to an NHK-22.2 top-layer signal with TpC (see FIGS. 7 and 10 to 15):

A fourth, complex example of an application of the subject invention, here according to FIGS. 10 and 11, to an NHK-22.2 top-layer signal, see FIG. 7, represents for the channels TpFL, TpFR, TpFC, TpC , TpBL, TpBR, TpSiL, TpSiR, TpBC the summation (the downmix)

TpFL <'> = TpFL + 0.5 * TpFC + 0.5 * TpSiL + 0.5 * TpC

TpFR <'> = TpFR + 0.5 * TpFC + 0.5 * TpSiR

TpBL <'> = TpBL + 0.5 * TpBC + 0.5 * TpSiL

TpBR <'> = TpBR + 0.5 * TpBC + 0.5 * TpSiR + 0.5 * TpC respectively

TpFL <'> = TpFL + 0.5 * TpFC + 0.5 * TpSiL TpFR <'> = TpFR + 0.5 * TpFC + 0.5 * TpSiR + 0.5 * TpC TpBL <'> = TpBL + 0.5 * TpBC + 0.5 * TpSiL + 0.5 * TpC TpBR <'> = TpBR + 0.5 * TpBC + 0.5 * TpSiR, where TpFL <'>, TpFR <'>, TpBL <'>, TpBR <'> again the vertices of the circumscribed square of FIG. 10 according to FIG. 3 correspond. It is now possible for each side of the square to perform a correlation comparison according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, in the manner described for the previous NHK 22.2 arrangements, and there are therefore the same signals as described above with the exception of new

TpFL + 0.5 * TpC = (TpFL + 0.5 * TpC + 0.5 * TpSiL) + ^ - 0.5 * (TpSiL - 2 * Δ4) = (TpFL + 0.5 * TpC + 0.5 * TpFC)

+ Δ4-0.5 * (TpFC - 2 * Δ ^ and

TpBR + 0.5 * TpC = (TpBR + 0.5 * TpC + 0.5 * TpBC) + Δ2- 0.5 * (TpBC - 2 * Δ3) 5 (TpBR + 0.5 * TpC + 0.5 * TpSiR)

+ Δ3- 0.5 * (TpSiR - 2 * Δ2) or

TpFR + 0.5 * TpC = (TpFR + 0.5 * TpC + 0.5 * TpSiR) + ^ - 0.5 * (TpSiR - 2 * Δ2) = (TpFR + 0.5 * TpC + 0.5 * TpFC) + Δ2- 0.5 * (TpFC - 2 * Δ ^

TpBL + 0.5 * TpC = (TpBL + 0.5 * TpC + 0.5 * TpBC) + Δ4- 0.5 * (TpBC - 2 * Δ3) = (TpBL + 0.5 * TpC + 0.5 * TpSiL) + Δ3 - 0.5 * (TpSiL - 2 * Δ4)

Then applies to the correlation comparison according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, the approximately obtained signals for TpFL + 0.5 * TpC and TpBR + 0.5 * TpC or TpFR + 0.5 * TpC and TpBL + 0.5 * TpC and for FIG. 3, that for the approximate signals resulting from adjacent correlation comparisons, only the difference η4-η3 or η2-η-ι or η-ι-η4 or η3-η2-immediate

15 influence after residual correction by the mean value of the sum .DELTA.-1 + .DELTA.2 + .DELTA.3 + .DELTA.4 has on the Residual resulting from this new correlation comparison.

Close would also be the downmix TpFL <'> = TpFL + 0.5 * TpFC + 0.5 * TpSiL + 0.25 * TpC TpFR <'> = TpFR + 0.5 * TpFC + 0.5 * TpSiR + 0.25 * TpC TpBL <'> = TpBL + 0.5 * TpBC + 0.5 * TpSiL + 0.25 * TpC TpBR <'> = TpBR + 0.5 * TpBC + 0.5 * TpSiR + 0.25 * TpC

Now consider

Δ2- Δ-, = η3- η4 = a-ι

Δ3 - Δ2 = η4 - τ | ι = a2

Δ4 - Δ3 = ηι - η2 = a3

Δι - Δ4 = η2 - η3 = a4and accordingly

(Δ2-Δ-ι) + (Δ3-Δ2) = η3-τ | ι = a-1 + a2

(Δ3 - Δ2) + (Δ4 - Δ3) = η4 - η2 = a2 + a3

(Δ4 - Δ3) + (Δι - Δ4) = τ | ι - η3 = a3 + a4

(Δι - Δ4) + (Δ2 - Δι) = η2 - η4 = a4 + ai

The same consideration as in the disclosure of the invention leads to T | 2 <=> | 4 + (^ 4 + ai) or τ | 2 = τ | 4 (a2 + a3) and ηι = η3 + (a3 + 84) or τ) ι = η3 - (ai + a2) which simply means that in this case the extraction of TpC can not be assigned a common residual.

Such a downmix is therefore possible, but not recommended, unless a residual associated with the extraction of the TpC is also transferred.

An alternative to the approximate extraction of the TpC by means of correlation comparison is the downmix

TpFL <'> = TpFL + (0.5 * TpFC + 0.5 * TpC) + 0.5 * TpSiL

TpFR <'> = TpFR + (0.5 * TpFC + 0.5 * TpC) + 0.5 * TpSiR

TpBL <'> = TpBL + 0.5 * TpBC + 0.5 * TpSiL

TpBR <'> = TpBR + 0.5 * TpBC + 0.5 * TpSiR, where for the correlation comparison according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, between TpFL <'> and TpFR <'> immediately (0.5 * TpFC + 0.5 * TpC - 2 * Δ-1) is extracted, and then a residual correction in the form described above can be carried out.

In fact, a localization between TpFC and TpC is psychoacoustically fraught with much fuzziness, which can be used selectively:

Instead of a correlation comparison for the extraction of the TpC, the imaging direction or the image width of the exact or approximated signal (0.5 * TpFC + 0.5 * TpC) is influenced by means of simple or dual panning known from the prior art so that it matches the original signal as much as possible comes immediately, and thus creates a psychoacoustically comparable impression to the original signal. Instead of a spatial coding or a correlation comparison for obtaining the TpC according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, only the parameters of the simple or dual panning are transmitted.

If other spatial encodings and decoders are to be used, such as the so-called inverse encoding, see above, these can be directly integrated into the above considerations:

For example, TpFC and TpC can be advantageously expressed by a non-linear inverse coding according to FIG. 9, as already explained above, which can also be coupled with simple or dual panning. The result is due to the psychoacoustic conditions a precise, natural hearing.

TpFL <'>, TpFR <'>, TpBL <'>, TpBR <'> and, if appropriate, the mean value A of all residuals Δ4, Δ2, Δ3, Δ4 (and possibly one from a correlation comparison, this according to the rules for Obtaining the real and imaginary parts of signals, see disclosure of the invention, for determining TpC resulting residual, or even the TpC signal itself) can be based on a base audio codec, which can be specially adapted for this purpose ("tuning") for efficient storage or transmission, see FIG. 6 (where in the present example additional spatial encoding and decoding can take place, for example, using the so-called inverse coding or else a simple or dual panning), for example in an encoder, and then for example, in a decoder, decompress.

Likewise, the systems described in total can make use of a Fourier transformation already carried out in the base audio coder, whereby the overall computation effort required can be significantly reduced.

16 Example Design of an Encoder and Decoder for a NHK 22.2 Top Layer Signal without TpC (See FIGS. 7 and 10 to 15):

Overall, the parameters associated with the described coding, see FIG. 6, can be transmitted as header information, as a data pulse or as a permanent data stream, for example from an encoder to a decoder.

A possible construction of an encoder and decoder for encoding and decoding an NHK 22.2 top layer signal without TpC is shown in FIGS. 12 to 14:

FIG. 12 shows an encoder module E, in which three adjacent input channels I * (t), ci * (t) and r * (t) or optionally a further input channel c '<1> * (t) or another input channel c, 2 * (t) are supplied. From these three input channels, a downmix Li <'> (t) = li * (t) + 0.5 * Ci * (t) and

Ri <'> (t) = r * (t) + 0.5 * Ci * (t) or

Li <'> (t) = li * (t) + 0.5 * Ci * (t) + 0.5 * ch * (t) and

Ri <'> (t) = r * (t) + 0.5 * Ci * (t) + 0.5 * ci2 * (t) is calculated, where appropriate Ci1 * (t) or Ci2 * (t) the nearest, not both downmix channels Lj <'> (t) and R, <'> (t) are mixed center channel (according to TpFC or TpSiR or TpBC or TpSiL). For both downmix channels L, <'> (t) and R, <'> (t), a Fast Fourier Transform (FFT) is then executed. On the one hand, these result in the output channels of the encoder module, and on the other hand, a correlation comparison according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, is applied to them. Similarly, a Fast Fourier Transform (FFT) is also performed for the input channel C * (t). It is now the residual Ai according to the formula

Δι (k) = * [Ci (k) -Ci (k)], which also represents an output signal of the encoder module. (The system described can be modified in accordance with disclosure of the invention by adding further input signals, for example based on FIG. 14, such that the residual Ai can also be expressed according to the formulas Δ = Li-Lj * or

Δ = Rj - Rj * can be calculated.)

FIG. 13 now shows the overall structure of the encoder. Four encoder modules E4, E2, E3, E4 are assigned the following input signals:

TpFC = Ci <*> (t)

TpFL = li <*> (t) = r4 <*> (t)

TpFR = rx '(t) = 12 <*> (t)

TpSiR = c2 <*> (t) = Ci2 <*> (t) = c3i <*> (t)

TpBC = c3 <*> (t)

TpBR = r2 <*> (t) = l3 <*> (t)

TpBL = r3 <*> (t) = l4 <*> (t)

TpSiL = c4 <*> (t) = Cn <*> (t) = c32 <*> (t)

The encoder module E-ι supplies the output signals LC (k), R- (k), Δ-ι (k). The encoder module E2 supplies the output signal A2 (k). The encoder module E3 supplies the output signals L3 <'> (k), R3 <'> (k), A3 (k). The encoder module E4 delivers the output signal A4 (k).

While the output signals L / (k), R / (k) and L3 <'> (k), R3 <'> (k) at the same time represent output signals of the encoder, the mean value A (k) of the residual Δ is finally -I (k), A2 (k), A3 (k), A4 (k). This average also represents an output signal of the encoder.

Fig. 14 now shows the structure of the decoder:

There is in this a first correlation comparison according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, based on the left input signal L (k) and the right input signal R / (k) instead, where only Ci (k) is calculated.

A second correlation comparison according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, takes place on the basis of the left input signal R- (k) and the right input signal L3 <'> (k), where both C2 (k) and L2 (k) are calculated.

A third correlation comparison according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, is found on the basis of the left input signal L3 <'> (k) and the right input signal R3 <'> (k). instead of calculating both C3 (k) and L3 (k).

A fourth correlation comparison according to the rules for obtaining the real and imaginary parts of signals, see disclosure of the invention, is found on the basis of the left input signal r3 <'> (k) and the right input signal L-1 <'> ( k), where both C4 (k) and L4 (k) and R4 (k) are calculated.

Based on the input signal for the frequency-dependent residual A (k), the following frequency-dependent channels are now calculated:

Ci (k) + 2Δ (k) = ci (k)

C2 (k) + 2Δ (k) Ξ c2 <*> (k)

L2 (k) - CX (k) - 2A (t) s 12 <*> (k)

C3 (k) + 2Δ (k) s c3 <*> (k)

L3 (k) - C2 (k) - 2A (k) s 13 <*> (k)

C4 (k) + 2Δ (k) s c4 <*> (k)

L4 (k) - C3 (k) - 2Δ (k) Ξ l4 <*> (k)

R4 (k) - Ci (k) - 2Δ (k) Ξ r4 <*> (k)

An inverse Fast Fourier Transform (IFFT) is now applied to each of these frequency-dependent channels.

Thus, the following output signals for the decoder, which approximately represent the identically named input signals of the encoder, result in detail:

[0118]

Ci (t) + 2A (t) = Ci <*> (t) = TpFC c2 (t) + 2A (t) = c2 <*> (t) = TpSiR 12 (t) - ci (t) - 2A t) = l2 <*> (t) = TpFR c3 (t) + 2Δ (t) £ = c3 <*> (t) = TpBC l3 (t) - c2 (t) - 2Δ (t) S l3 <* > (t) = TpBR c4 (t) + 2Δ (t) S c4 <*> (t) = TpSiL l4 (t) - c3 (t) - 2Δ (t) = l4 <*> (t) = TpBL r4 (t) - Ci (t) - 2Δ (t) = r4 <*> (t) = TpFL

Closing remarks:

Overall illustrated principles are algorithmically arbitrarily expandable and thus allow overall the efficient compression of multi-signals of any, even highest, order based on a downmix, this for the purpose of efficient storage or transmission, for example between an encoder and a decoder.

Claims 9 to 42 use the method according to Claims 1 to 8 for determining at least one common signal and / or a first individual signal and / or a second individual signal from two input signals.

18

Claims (42)

Signals. Alternatively, claims 9 to 42 could use any other method for determining a common signal, a first individual signal and a second individual signal from two input signals. Furthermore, the storage and / or transmission of data (eg a file or other storage means or transmission means) with a Downmixsignal and / or with a residual from a plurality of residuals Residual and / or with a Panningparameterset and / or with parameters of a inverse encoding disclosed. A multichannel signal with n channels can in turn contain a further multichannel signal with n-1> 2 channels, a further multichannel signal with n-2> 2 channels, etc. Conversely, a further multi-channel signal of higher order can be derived from a multi-signal with n> 2 or n-1> 2 or n-2> 2 etc. channels. claims A method for extracting at least one output signal from two input signals; characterized by providing first frequency dependent input signal components (Lj <'> (k)) and second frequency dependent input signal components (Ri' '> (k)) for a plurality of frequencies; Comparing the signs of the first frequency dependent input signal component (k)) and the second frequency dependent input signal component (Rj <'> (k)) of a frequency (k) of the plurality of frequencies; Determining at least one of a first frequency-dependent individual signal component (Lj (k)) of a first individual signal, a second frequency-dependent individual signal component (R, (k)) of a second individual signal and a common frequency-dependent signal component (C, (k)) of the frequency (k) Plurality of frequencies based on the sign comparison; Determining the at least one output signal on the basis of the first frequency-dependent individual signal components (Lj (k)) of the plurality of frequencies and / or the second frequency-dependent individual signal components (R, (k)) of the plurality of frequencies and / or the common frequency-dependent signal components (Cj (k)) of the plurality of frequencies. 2. The method according to claim 1, wherein the step of determining at least one of the first frequency-dependent individual signal component (L, (k)), the second frequency-dependent individual signal component (R, (k)) and the common frequency-dependent signal component (C, (k)) the frequency (k) has the following steps: For identical sign of the first and second frequency-dependent input signal component (Lj <'> (k), Rj <'> (k)) of the frequency (k): Determination of the common frequency-dependent signal component (C, (k)) of the frequency k on the basis the smaller amount of the first and second frequency-dependent input signal components (L, <'> (k), R, <'> (k)) of the frequency (k); and or With an identical sign of the first and second frequency-dependent input signal component (k), R, k (k)) of the frequency (k): If the first frequency-dependent input signal component (L, k (k)) is greater in magnitude than the second frequency-dependent one Input signal component (Rf (k)), determination of the first frequency-dependent individual signal component (L, (k)) of the frequency (k) on the basis of the difference of the first frequency-dependent input signal component (L, <'> (k)) from the second frequency-dependent input signal component ( rj < '> (k)); For nonidentical signs of the first and second frequency-dependent input signal components (Lj <'> (k), Rj <'> (k)) of the frequency (k): determination of the first frequency-dependent individual signal component (Lj (k)) of the frequency (k) the base of the first frequency-dependent input signal component (Lj <'> (k)); and or With identical sign of the first and second frequency-dependent input signal component (Lj <'> (k), Rj <'> (k)) of the frequency (k): If the first frequency-dependent input signal component (Lj <'> (k)) is smaller than the second frequency-dependent input signal component (Rj <'> (k)), determining the second frequency-dependent individual signal component (Rj (k)) of the frequency (k) on the basis of the difference of the second frequency-dependent input signal component (Rj <'> (k)) of the first frequency-dependent input signal component (Lj <'> (k)); For non-identical sign of the first and second frequency-dependent input signal components (Lj <'> (k), Rj <'> (k)) of the frequency (k): determination of the second frequency-dependent individual signal component (R, (k)) of the frequency (k) on the basis of the second frequency-dependent input signal component (R, k (k)). 3. The method of claim 2, wherein the step of determining at least one of the first frequency-dependent individual signal component (Lj (k)), the second frequency-dependent individual signal component (R, (k)) and the common frequency-dependent signal component (Cj (k)) of the frequency (k) comprises the following steps: With identical sign of the first and second frequency-dependent input signal component (Lj <'> (k), Rj <'> (k)) of the frequency (k): determination of the smaller amount of the first and second frequency-dependent input signal component (Lj <'> (k) , Rj <'> (k)) of the frequency as a common frequency-dependent signal component (C, (k)) of the frequency k; If the sign of the first and second frequency-dependent input signal components (Lj <'> (k), Rj <'> (k)) of the frequency (k) is not identical, setting the common frequency-dependent signal component (Cj (k)) of the frequency (k) to zero; and / or With identical sign of the first and second frequency-dependent input signal component (Lj <'> (k), Rj <'> (k)) of the frequency (k): If the first frequency-dependent input signal component (Lj <'> (k)) in terms of magnitude is greater than the second frequency-dependent input signal component (Rj <'> (k)), determination of the first frequency-dependent individual signal component (Lj (k)) of the frequency (k) as the difference of the first frequency-dependent input signal component (Lj <'> (k)) of 19 the second frequency-dependent input signal component (FV (k)), otherwise determining the first frequency-dependent individual signal component (L, (k)) of the frequency (k) as zero; If the sign of the first and second frequency-dependent input signal components (k), R, k (k) is not identical, the frequency k is determined as the first frequency-dependent input signal component (L, (k)); and or If the sign of the first and second frequency-dependent input signal component (L, k (k), R ± k (k)) of the frequency (k) is identical, the amount of the first frequency-dependent input signal component (L, k (k)) is absolute is smaller than the second frequency-dependent input signal component (R, <'> (k)), determination of the second frequency-dependent individual signal component (R, (k)) of the frequency (k) as the difference of the second frequency-dependent input signal component (R, k (k)) from the first frequency-dependent input signal component (k)), otherwise determining the second frequency-dependent individual signal component (R, (k)) of the frequency (k) as zero; In the case of a not identical sign of the first and second frequency-dependent input signal components (L, k (k), R, k (k)) of the frequency (k): determination of the second frequency-dependent individual signal component (R, (k)) of the frequency ( k) as a second frequency-dependent input signal component (R, <'> (k)). 4. Method according to one of claims 1 to 3, wherein the first and second frequency-dependent input signal components (Rf (k), L, <'> (k)) are complex-valued and the step of determining at least one of the first frequency-dependent individual signal component (L, (k)), the second frequency-dependent individual signal component (R, (k)) and the common frequency-dependent signal component (C, (k)) of the frequency (k) are performed separately once for the real part and / or once for the imaginary part. 5. The method of claim 1, wherein providing first frequency-dependent input signal components (L <'> (k)) and second frequency-dependent input signal components (R <'> (k)) Fourier transforms the first input signal from the time domain into the frequency domain and of the second input signal from the time domain to the frequency domain. 6. The method according to any one of claims 1 to 5, wherein the at least one output signal consists of frequency-dependent output signal components. 7. The method according to any one of claims 1 to 5, wherein the at least one output signal by inverse Fourier transform of on the basis of the first frequency-dependent individual signal components (L, (k)) a plurality of frequencies and / or the second frequency-dependent individual signal components (R, ( k)) a plurality of frequencies and / or the common frequency-dependent signal components (C, (k)) of a plurality of frequencies formed frequency-dependent signal components is formed. 8. The method according to any one of claims 1 to 7, wherein the step of comparing the sign of the frequency and determining at least one of a first frequency-dependent individual signal component (L, (k)) of a first individual signal, a second frequency-dependent individual signal component (R, (k )) of a second individual signal and a common frequency-dependent signal component (C, (k)) of the frequency (k) is performed on the basis of the sign comparison for the plurality of frequencies, respectively. 9. A method of encoding three channels of a multi-channel signal into at least one channel of a downmix signal, comprising the following steps: Downmixing three channels of the multi-channel signal into two channels of the downmix signal; Estimating at least one of the three channels by the method of any one of claims 1 to 8 with the two channels of the downmix signal as inputs; Determining the residual by a difference of an original channel of the multi-channel signal and the estimated channel of the multi-channel signal. 10. The method of claim 9, wherein the residual is determined by determining a frequency-dependent residual component of a frequency (k) by the difference of a frequency-dependent signal component of the original channel of the multi-channel signal estimated from this frequency (k) and the estimated frequency-dependent signal component of that frequency (k) is determined, wherein the residual is determined on the basis of the frequency-dependent residual components of a plurality of frequencies. 11. The method according to claim 10, further comprising determining the frequency-dependent signal components of the two channels of the downmix signal by Fourier transformation of the two channels of the downmix signal and / or determining the frequency-dependent signal components of the original channel to be estimated of the multichannel signal by Fourier transformation of this original channel of the multichannel signal to be estimated. 12. The method according to claim 1 1, wherein the frequency-dependent signal components of at least one of the two channels of the downmix signal from a base audio encoder are used. 13. The method according to any one of claims 9 to 12, wherein at least one channel of the downmix signal is compressed by a base audio encoder. 14. The method according to any one of claims 9 to 13, wherein the three channels of the multi-channel signal are three adjacent channels of the multi-channel signal. 20 15. A method of encoding a multi-channel signal with n channels into a downmix signal with m channels, with m <n comprising the following steps: Encoding three first channels of the multi-channel signal and determining a first residual according to the method of any one of claims 9 to 14; Encoding three second channels of the multi-channel signal and determining a second residual according to the method of any one of claims 9 to 14; Determining an averaged residual on the basis of the first residual and the second residual; Outputting the channels of the downmix signal determined by the encoding, and the averaged residual. 16. The method of claim 15, wherein the three first channels of the multi-channel signal correspond to a first channel, a second channel and a third channel of the multi-channel signal, wherein the three second channels of the multi-channel signal corresponding to the third channel, a fourth channel and a fifth channel of the multi-channel signal wherein the second channel is associated with a position between the positions associated with the first and third channels, and the fourth channel is associated with a position between the positions associated with the third and fifth channels, wherein the encoding of the three first channels of the multi-channel signal after The method of claim 9, further comprising determining a channel of the downmix signal based on the second channel, the third channel, and the fourth channel of the multi-channel signal. 17. The method according to claim 15 or 16, wherein the three first channels and the three second channels lie in a horizontal plane. 18. The method according to any one of claims 15 to 17, further comprising: Encoding three third channels of the multi-channel signal and determining a third residual according to the method of any one of claims 9 to 14; Encoding three fourth channels of the multi-channel signal and determining a fourth residual according to the method of any one of claims 9 to 14; Determining an averaged residual on the basis of the first, second, third and fourth residuals; Outputting the particular channels of the downmix signal, and the averaged residual. 19. The method of claim 18, wherein the multi-channel signal comprises at least eight channels of a horizontal plane having a front left channel, a front center channel, a front right channel, a right side channel, a rear right channel, a rear center channel, a back left Channel and a left side channel, and wherein the three first channels correspond to the front left channel, the front center channel and the front right channel, the three second channels correspond to the front right channel, the right side channel and the rear right channel, the three third channels correspond to the rear right channel, the rear center channel, and the rear left channel, and the three fourth channels correspond to the rear left channel, the left side channel, and the front left channel, and four of the eight are designated in the four encodings Channels of the downmix signal are output. The method of claim 18 or 19, wherein the multi-channel signal comprises a ninth channel located in the middle of the eight channels or adjacent to one of these eight channels, and wherein the ninth channel is integrated into one of the eight channels before the four channels of the downmix signal. 21. The method of claim 20, wherein the multi-channel signal has a ninth channel that is in the middle of the eight channels or adjacent to one of these eight channels, and wherein the ninth channel is added to one of the eight channels before the four channels of the downmix signal be determined. The method of claim 21, wherein a panning parameter set for the ninth channel and the channel added to the ninth channel is determined, and the panning parameter set is also output. 23. The method according to any one of claims 15 to 22, wherein one of the three first channels of the multi-channel signal are combined prior to encoding with another channel of the multi-channel signal to a common channel, and the parameters for separating this common channel with output. 24. The method of claim 23, wherein the parameters for separating this common channel are based on at least four parameters: an angle between a sound source and a microphone main axis, a certain left opening angle, a certain right opening angle and a directional characteristic of the common signal. 25. A method for decoding a multi-channel signal with n channels from a downmix signal with m channels, with m <n comprising the following steps: Providing m channels of the downmix signal; Determining p> m channels by at least one application of the method according to any one of claims 1 to 8, each having two channels of the downmix signal as input signals. 26. The method of claim 25, comprising providing a residual (A); Correcting at least one channel of the multi-channel signal by the Residual (A). 21 27. The method of claim 26, wherein at least one channel of the multi-channel signal is based on at least one of the first individual signal (L, (k)), the second individual signal (R, (k)) and the common signal (C, (k) ) determined by the method according to claims 1 to 8, and wherein the first individual signal (Lj (k)) is corrected by subtraction of a residual and / or the second individual signal (R, (k) ) is corrected by subtracting the residual and / or the common signal (Cj (k)) by adding the double residual. 28. The method according to any one of claims 25 to 27, wherein at least a first channel, a second channel and a third channel of the downmix signal in at least a first channel, a second channel, a third channel, a fourth channel and a fifth channel of the multi-channel signal following steps are decoded: Determining the second channel of the multi-channel signal based on the common signal of the first channel and the second channel of the downmix signal determined according to the method of claims 1 to 8; and determining the fourth channel of the multi-channel signal based on the common signal of the second channel and the third channel of the downmix signal determined according to the method of claims 1 to 8. 29. The method of claim 28, wherein an averaged residual is received, the second channel of the multi-channel signal is determined based on the common signal of the first channel and the second channel of the downmix signal and the averaged residual, and the fourth channel of the multi-channel signal on the base of the second channel and the third channel of the downmix signal and the averaged residual. 30. The method of claim 28 or 29, wherein the third channel of the multi-channel signal is determined based on the common signal of the first channel and the second channel and the common signal of the second channel and the third channel. The method of any of claims 25 to 30, comprising receiving four channels of the downmix signal and an average residual (Δ); Determining at least eight channels of the multi-channel signal by applying the method of any one of claims 1 to 8 four times with the four combinations of two channels of the downmix signal as inputs; correcting the eight specific channels of the multi-channel signal by the provided averaged residual (Δ). 32. The method of claim 31, further comprising separating a signal component of one of the eight channels for a ninth channel of the multi-channel signal. 33. The method according to claim 32, wherein parameters for separating the ninth channel of the multi-channel signal are received, the ninth channel is separated on the basis of these parameters. 34. The method of claim 32, wherein the ninth channel of the multi-channel signal is in the middle of the eight channels, or adjacent one of the eight channels of the multi-channel signal. 35. The method according to claim 32, wherein the separation is based on provided parameters for the panning and / or an inverse coding and / or a defined relationship. 36. A computer program configured, when executed on a processor, to carry out the method steps of one of claims 1 to 35. 37. Apparatus for extracting at least one output signal from two input signals; characterized by receiving means for receiving first frequency dependent input signal components (L, <'> (k)) and second frequency dependent input signal components (Rf (k)) for a plurality of frequencies; a comparison device for comparing the signs of the first frequency-dependent input signal component (k)) and the second frequency-dependent input signal component (R, <'> (k)) of a frequency (k) of the plurality of frequencies; a calculating means for determining at least one of a first frequency-dependent individual signal component (L, (k)) of a first individual signal, a second frequency-dependent individual signal component (R, (k)) of a second individual signal and a common frequency-dependent signal component (C, (k)) of the frequency (k) the plurality of frequencies based on the sign comparison; and the calculating device is further configured to determine the at least one output signal on the basis of the first frequency-dependent individual signal components (L, (k)) of the plurality of frequencies and / or the second frequency-dependent individual signal components (R, (k)) of the plurality of frequencies and / or the common frequency-dependent signal components (C, (k)) of the plurality of frequencies. 38. An encoding device for encoding three channels of a multi-channel signal into at least one channel of a downmix signal comprising: Downmixer for downmixing three channels of the multi-channel signal into two channels of the downmix signal; Apparatus according to claim 36, including the two channels of the downmix signal from the downmixer as inputs; Residual device for determining the residual by a difference of an original channel of the multi-channel signal and the estimated channel of the multi-channel signal. 39. An encoding device for encoding a multi-channel signal with n channels into a downmix signal with m channels, with m <n, comprising: 22, an encoding device according to claim 37 or 38 for encoding three first channels of the multi-channel signal and determining a first residual; an encoding device according to claim 37 or 38 for encoding three second channels of the multi-channel signal and determining a second residual; An averaging means for determining an averaged residual on the basis of the first residual and the second residual; Output device for outputting the channels of the downmix signal and the averaged residual determined by the encoding. 40. The encoding device of claim 38, further comprising: an encoding device according to claim 37 for encoding three third channels of the multi-channel signal and determining a third residual; an encoding device according to claim 37 for encoding three fourth channels of the multi-channel signal and determining a fourth residual; wherein the averaging device is configured to determine the averaged residual based on the first, second, third and fourth residuals. 41. A decoding device for decoding a multi-channel signal with n channels from a downmix signal with m channels, with m <n, comprising: a receiving device for receiving m channels of the downmix signal; at least one device according to claim 36, each having two channels of the downmix signal as input signals for determining p> m channels of the multi-channel signal. 42. System comprising: An encoding device according to claim 38 or 39 for encoding a multichannel signal having n channels into a downmix signal having m channels; Transmission means for transmitting the m channels of the downmix signal; and Decoding device according to claim 40 for decoding the m channels of the downmix signal into p> m channels of the multi-channel signal. 23