TWI458365B - Apparatus and method for generating a level parameter, apparatus and method for generating a multi-channel representation and a storage media stored parameter representation - Google Patents

Description

Apparatus and method for generating level parameters, apparatus and method for generating multi-channel representation, and storage medium for storing parameter representation

The present invention relates to encoding of a multi-channel representation of an audio signal using spatial parameters. The present invention teaches a new method for estimating and defining appropriate parameters that can be used to reconstruct a multi-channel signal from some channels (less than the number of output channels). In particular, emphasis is placed on minimizing the bit rate of the multi-channel representation and providing one of the multi-channel signals encoded representations that can easily encode and decode the material for all possible channel configurations.

PCT/SE02/01372, entitled "Efficient and Adjustable Parametric Stereo Coding for Low Bit Rate Audio Coding Applications", has been shown to reconstruct a stereo image very similar to the original stereo image from a mono signal ( Assume that there is a very tight representation of the stereo image). The basic principle is to divide the input signal into frequency bands and time segments, and to estimate inter-channel intensity difference (IID) and inter-channel coherence (ICC) for these bands and time segments. The first parameter is a measure of the power distribution between two channels in a particular frequency band, and the second parameter is an estimate of the correlation between the two channels of that particular frequency band. On the decoder side, by assigning the mono signal between the two output channels in accordance with the IID data and by adding a decorrelated signal to maintain the channel correlation of the original stereo channel, from the mono channel The signal reconstructs the stereo image.

For a multi-channel case (multiple channels in the context represent more than two output channels), several additional issues must be accounted for. There are now several multi-channel configurations. The most commonly known is the 5.1 configuration (middle channel, front left/right channel, surround left/right channel, and LEE channel). However, there are many other configurations now. From a complete encoder/decoder system perspective, it is desirable to have one that can use the same set of parameters (eg, IID and ICC) or a subset thereof for all channel configurations. ITU-R BS.775 defines several down-mix schemes to enable a channel configuration that includes fewer channels from a particular channel configuration. Instead of having to decode all of the channels and relying on the next line of mixing, it is desirable to have a multi-channel representation that enables a receiver to retrieve parameters relating to the configuration of the hand channel before decoding the channels. Furthermore, it is desirable from an adjustable or inline coding point of view to have an inherently adjustable set of parameters, in which, for example, data corresponding to the surround channels can be stored in one of the bitstream enhancement layers. .

Contrary to the above, it may also be desirable to use different parameter definitions depending on the characteristics of the signal being processed in order to switch between parameterizations that would result in the lowest bit rate burden for the current signal segment being processed.

Another representation of a multi-channel signal using a summed or downmix signal and additional parameter additional information is known as Binaural Cue Coding (BCC). This technique is described in the November 2011 issue of the 6th issue of the IEEE Speech Processing Journal by the authors F. Baumgarte and C. Faller, "Two-Channel Signal Coding - Part I: The Foundation and Design Principles of Auditory Psychology" And the author of the IEEE Speech Processing Journal, Vol. 11, No. 6, November 2003, is the "Two-Channel Signal Coding - Part Two: Architecture and Applications" by C. Faller and F. Baumgarte.

In general, two-channel signal coding is a method of multi-channel spatial representation based on the next line of mixing channels and additional information. Several parameters calculated by a BCC encoder for audio reconstruction or audio and used by a BCC decoder include inter-channel level differences, inter-channel time differences, and inter-channel coherence parameters. These inter-channel signals are the determining factor for the perception of a spatial image. These parameters are the blocks provided to the time samples of the original multi-channel signal and are also provided with frequency selectivity such that each block of the multi-channel signal samples has several signals for several frequency bands. In the general case of C-playing channels, each of the sub-bands between the complex pairs of channels (i.e., for each channel relative to a reference channel) takes into account the level differences of the channels and such The time difference between channels. One channel is defined as a reference channel for the level difference between each channel. Due to the level difference between the channels and the time difference between the channels, it is possible to provide any direction from one source to one of the plurality of pairs of speakers of a used playback device. In order to determine the width of a given source of diffusion, it is sufficient to consider a parameter for each subband of all audio channels. This parameter is the coherence parameter between the channels. The width of the supplied sound source is controlled by modifying the sub-band signals such that all possible channel pairs have the same inter-channel co-modulation parameters.

In BCC encoding, the level difference between all channels between the reference channel 1 and any other channel is determined. When, for example, determining that the center channel is the reference channel, calculating a level difference between the first channel between the left channel and the center channel, and a second between the right channel and the center channel Level difference between channels, level difference between the left channel between the left surround channel and the center channel, and level difference between the fourth channel between the right surround channel and the center channel . This episode describes a 5-channel architecture. When the 5-channel architecture additionally includes a low frequency enhanced channel (also known as a "sub-woofer" channel), the low frequency enhanced channel and the center channel are calculated The fifth channel level difference between (the single reference channel).

When using this single downmix channel (also known as "single" channel) and transmitting signals such as ICLD (inter-channel potential difference, ICTD (inter-channel time difference) and ICC (channel-to-channel homology)) These signals are used to modify the spectral coefficients of the single signal during the original multi-channel. A positive real number is used to determine the level modification of each spectral coefficient to implement the level modification. A complex number is used to determine the phase modification of each spectral coefficient to produce the inter-channel time difference. Another feature determines the coherence effect. The factor of the level modification of each channel is calculated by first calculating the factor of the reference channel. The factor of the reference channel is calculated such that the sum of the powers of all channels is the same as the power of the combined signal for each frequency portion. Then, based on the level modification factor of the reference channel, the individual ICLD parameters are used to calculate the level modification factors of the other channels.

Therefore, in order to perform BCC synthesis, the level modification factor of the reference channel is calculated. For this calculation, all ICLD parameters for a band are required. Then, based on the level modification of the mono, the level modification factor of the other channels (i.e., the channels other than the reference channel) can be calculated.

The disadvantage of this method is that for a complete reconstruction, the level difference between each channel is required. This requirement creates a greater problem when an error-prone transmission channel occurs. Since each level difference between channels is required to calculate each multi-channel output signal, each error within the inter-channel level difference will result in an error in the reconstructed multi-channel signal. In another case, although the level difference between the one channels is only required for, for example, the left surround channel or the right surround channel, the reconstruction cannot be performed when the inter-channel level difference is lost during transmission. The left surround channel and the right surround are included because the important information is included in the front left channel (hereinafter referred to as the left channel), the front stone channel (hereinafter referred to as the right channel), and the center channel. The channel is not important for multi-channel reconstruction. This situation gets worse when the inter-channel level difference of the low frequency enhanced channel is lost during the transmission. In this case, although the low frequency enhanced channel is not decisive for the listener's listening comfort, there may be no multi-channel reconstruction or only one erroneous multi-channel reconstruction. Because the error in the level difference between the single channels is propagated to the error in each reconstructed output channel.

The problem with parametric multi-channel representation is that inter-channel level differences (eg, ICLD in BCC encoding or balanced values in other parametric multi-channel representations) are typically provided as relative values rather than absolute values. In BCC, an ICLD parameter describes the level difference between a channel and a reference channel. A balance value can also be provided as a ratio between the two channels in a pair of channels. When the multi-channel signal is reconstructed, the level difference or balance parameter is applied to a basic channel, which may be a mono basic channel or a stereo basic sound with two basic channels. Signal. Thus, the energy contained in at least one of the basic channels is distributed along five or six reconstructed output channels. Thus, the absolute energy in a reconstructed output channel is determined by the inter-channel level difference or the balance parameter and the energy of the down-mix signal at the receiver input.

A level change occurs when the energy of the down-mix signal at the receiver input changes with respect to one of the down-mix signals output by an encoder. In this context, it is emphasized that when the parameters are frequency selective, depending on the parametric architecture used, this level change not only results in a general volume change of the established signal, but also a large number of artifacts. When, for example, a certain frequency band of the downlink mix signal is more frequently manipulated in the frequency range than in one of the other positions, since the frequency component in the output channel of the certain frequency band has a too low Or too high level, so this control is clearly visible in the reconstructed output signal.

In addition, changing the level operation in a timely manner will also cause the total level of the reconstructed output signal to change over time and is therefore considered an annoying artifact.

Although the above situation focuses on the level operation caused by encoding, transmitting and decoding the next line of mixed signals, other level shifts may occur. Due to the phase dependence between the different channels to be downmixed into one or two channels, the following occurs: the mono signal has a sum that is not equal to the energy in the original signal. Since the downmix is usually implemented, for example, by adding a time waveform to sample-wise, although the left and right signals of course have a certain signal energy, for example, a phase of 180 degrees between the two signals The difference will result in complete cancellation of the two channels in the downstream mix signal, which in turn results in zero energy. Although this situation will very unlikely to occur under normal conditions, energy changes will still occur because all signals are of course not completely uncorrelated. Since the energy of the reconstructed output signal will be different from the energy of the original multi-channel signal, such variations will also result in volume fluctuations in the reconstructed output signal and will also result in artifacts.

It is an object of the present invention to provide a parametric concept that can result in a multi-channel reconstruction with improved output quality.

The object is to generate a device for reconstructing a multi-channel representation according to a device for generating a level parameter according to item 1 of the patent application scope, according to claim 7 of the patent application scope, according to the scope of claim 9 A method for generating a level parameter, a method for generating a reconstructed multi-channel representation according to claim 10 of the patent application, a computer program according to claim 11 or a patent application scope A parameter of 12 items is expressed.

The present invention is based on the following findings: for high quality reconstruction and in view of the elastic coding/transmission and decoding architecture, an additional level parameter is transmitted along with a multi-channel signal down-mix signal or parameter representation such that A multi-channel reconstructor can use the level parameter together with the level difference parameter and the downmix signal to reproduce a multi-channel output signal without encountering a level change or a frequency selective level. Artificial factors caused.

In accordance with the present invention, the level parameter is calculated such that at least the energy of the next mixing channel is weighted (e.g., multiplied or divided) by the level parameter equal to the sum of the energy of the original channels.

In one embodiment, the level parameter is obtained from the ratio of the energy of the (equal) downstream mixing channel to the sum of the energy of the original channels. In this embodiment, any level difference between the (equal) downstream mixing channel and the original multi-channel signal is calculated on the encoder side and input to the data stream as a level correction. The factor, the level correction factor is considered to be an additional parameter that can also be provided to a block of the sample of the (equal) downmix channel and provided to a certain frequency band. Therefore, a new level parameter is added for each region and frequency band (there is an inter-channel level difference or balance parameter).

The present invention also provides flexibility because the present invention allows for the transmission of one of the multi-channel signals to be different from the downstream mix of the line mix under which the parameters are based. For example, a broadcast station does not wish to play a downlink mix signal generated by a multi-channel decoder, but it is desirable to play a downmix signal generated by a sound engineer in a studio (based on the subjective view of human beings) This happens when you create a mix of impressions. However, the player may also wish to transmit multi-channel parameters related to this "main downmix". According to the present invention, the level parameter is used to provide an adaptation between the parameter group and the main downmix, in which case the level parameter is the level between the main downmix and the parameter downmix. Poor, where the parameter group is downmixed according to the parameter.

An advantage of the present invention is that the additional level parameter is provided because the parameter set for the next line of mixing signal can also be adapted to another downstream mix, wherein the other downstream mix is not generated during parameter calculation. Improved output quality and improved flexibility.

For the reduction of the bit rate, it is preferable to apply the delta-encoding of the new level parameter as well as quantization and entropy coding. In particular, since the variation between inter-band or time blocks will not be so high that a relatively small difference can be obtained, the use of this concatenation followed by entropy coding (eg Huffman encoder) to allow good coding gain The possibility, so delta-encoding will result in a high coding gain.

In a preferred embodiment of the invention, a multi-channel signal parameter representation comprising at least two different balance parameters is used, the at least two different balance parameters representing a balance between two different channel pairs. In particular, the flexibility, adjustability, error resistance, and even bit rate rate efficiency are the result of the fact that the first channel pair (the basis of the first balance parameter) is different from the second channel pair (the second balance parameter) According to the basis, the four channels in which these pairs of channels are formed are different from each other.

Thus, the inventive concept differs from the single reference channel concept and the use of a multi-balance or over-balance concept that is more straightforward and more natural to the human voice. In particular, the pairs of channels that make up the first and second balance parameters may include some combination of the original channel, the downmix channel, or preferably the input channel.

A balance parameter obtained from the center channel (as the first channel) and the addition of the left original channel and the right original channel (as the second channel of the channel pair) have been found It is always useful to provide a precise energy distribution between the center channel and the left and right channels. It is noted that in this context these three channels usually comprise most of the information of the sound field, wherein in particular the left-right stereo localization is not only affected by the balance between the left and the right, but also by the central and left and right. The effect of the balance between the two. This observation is reflected by the use of this balancing parameter in accordance with a preferred embodiment of the present invention.

Preferably, when transmitting a single single downmix signal, it has been found that in addition to the center/left + right balance parameter, there is a left/right balance parameter, a back-left/back-right balance parameter, and a front The /after balance parameter is the best solution for one-bit rate-effective parameter representation, which is elastic, error-resistant and immune to large artificial factors.

On the receiver side, the multi-balance representation of the present invention is additionally used in the line mixing architecture to generate the downstream mixing channel, as compared to the BCC synthesis of each channel calculated by the transmitted information alone. Information on. Therefore, the data on the downstream mixing architecture (not used in the prior art system) is also used in addition to the balancing parameters. Therefore, the upstream mixing operation is implemented to determine the balance between the channels in a reconstructed multi-channel signal (forming a pair of channels for a balanced parameter) by the balance parameter.

This concept (i.e., different balance parameters have different pairs of channels) can produce some channels without knowing each transmission balance parameter. In particular, the left, right and center channels can be reconstructed without knowing any post-left/back-right balance or without knowing the front/back balance. This result allows very fine-tuning of the ability to adjust, since an additional parameter is extracted from a single stream or an additional balanced parameter is transmitted to a receiver, thus allowing reconstruction of one or more additional channels. This is in contrast to the prior art single reference system in which a level difference between channels is required to reconstruct all subgroups or only a subgroup of all reconstructed output channels.

The inventive concept is also flexible because the choice of the balance parameters can be adapted to a certain reconstruction environment. A front-back balance parameter allows calculation of the combined surround when, for example, a 5-channel setup forms the original multi-channel signal setup and when a 4-channel setup forms a reconstructed multi-channel setup The channel does not require any knowledge of the left surround channel and the left surround channel, wherein the reconstructed multi-channel device has only a single surround speaker, and the single surround speaker is, for example, disposed behind the listener. This is in contrast to a single reference channel system in which the inter-channel level difference of the left surround channel and the inter-channel level of the right surround channel must be retrieved from the stream. difference. Then, the left surround channel and the right surround channel must be calculated. Finally, two channels must be added to obtain the single surround speaker channel for a 4-channel reconstruction setup. Because the more intuitive and user-oriented balanced parameter representation is not limited to a single reference channel, the combination of the original channels can be allowed to be used as a balanced channel pair of channels and thus automatically The combined surround channel is sent, so it is not necessary to implement all of these steps in the balanced parameter representation.

The present invention is related to the parametric multi-channel representation of audio signals. The present invention provides an efficient way to define appropriate parameters for the multi-channel representation and also provides the ability to retrieve parameters for representing the desired channel configuration without decoding all channels. The present invention further addresses the problem of selecting an optimal parameter configuration for a particular signal segment in order to minimize the bit rate required to encode the spatial parameter for that particular signal segment. The present invention also outlines how to apply a decorrelation method that was previously only applicable to two channel situations in a general multi-channel environment.

In a preferred embodiment, the invention includes the following features:

- downmixing the multichannel signal to one or two channel representations on the encoder side;

- the multi-channel signal is known to define parameters for representing the multi-channel signals in order to minimize the bit rate in an elastic per-frame basis or to enable the decoder to capture the bit stream On-channel configuration;

- assuming that the channel configuration is currently supported by the decoder, the relevant parameter set is retrieved on the decoder side;

- assuming the current channel configuration, producing the required number of mutual decorrelation signals;

- assuming that the parameter set is decoded by the bit stream data and the decorrelated signals, reconstructing the output signals;

Defining the parameterization of the multi-channel audio signal so that the same parameter or a subset of the parameters can be used without regard to the channel configuration;

Defining the parameterization of the multi-channel audio signal so that the parameters can be used in an adjustable coding architecture at which a subset of the parameter set is transmitted in different layers of the adjustable stream;

Defining the parameterization of the multi-channel audio signal such that energy reconstruction from the output signal of the decoder is not compromised by an audio codec for encoding the downstream mix signal;

- switching between different parameterizations of the multi-channel audio signal in order to minimize the bit rate burden used to number the parameterization;

Defining a parameterization of the multi-channel audio signal, comprising a parameter indicating an energy correction factor of the downstream mix signal;

- using several de-correlated decorrelators to reconstruct the multi-channel signal;

- reconstructing the multi-channel signal from an upstream mixing matrix calculated from the set of transmission parameters.

The invention is now described by way of example with reference to the accompanying drawings, which are not intended to limit the scope or spirit of the invention.

The embodiments described below are merely illustrative of the principles of multi-channel representation of the present invention in audio signals. It will be appreciated that modifications and variations of the configurations and details described herein will be apparent to those skilled in the art. Therefore, the meaning of the invention is to be limited only by the specific details of the description and description of the embodiments herein.

In the following description of the present invention, which outlines how to parameterize IID and ICC parameters and how to apply these parameters in order to reconstruct a multi-channel representation of an audio signal, assume that all mentioned signals are sub-band signals or corresponding sounds in a filter stop. Some other frequency selective representation of a portion of the entire frequency range of the track. Thus, it is understood that the present invention is not limited to a particular filter bank, and that a frequency band is shown below for a sub-band representation of the signal, and that the same operation is applied to all sub-band signals.

Although a balance parameter is also referred to as an Inter-Channel Intensity Difference (IID) parameter, it is not necessary to emphasize the energy or intensity in the first channel of the channel pair. And the energy or intensity of the second channel in the pair of channels. Typically, the balance parameter represents the localization of a sound source between the two channels of the pair of channels. While this localization is typically provided by energy/level/intensity differences, other characteristics of the signal (eg, power measurements of two channels or time or frequency encapsulation of the channels, etc.) may be used.

In Fig. 1, a different channel of a 5.1 channel configuration is presented, where a(t) 101 represents the left surround channel, b(t) 102 represents the left front channel, and c(t) 103 represents the center. The channel, d(t) 104 represents the right front channel, e(t) 105 represents the right surround channel, and f(t) 106 represents the LEF (low frequency effect) channel.

Suppose we define the expectation operator as:

And therefore the energy of the channel outlined above can be defined as follows (here the left surround channel is used as an example):

A =

The 5-channel downmix is made on the encoder side as a 2-channel representation or a 1-channel representation. This can be done in several ways, and one commonly used way is the ITU Downmix as defined below: 5.1 to 2-channel downmix:

l _d =αb(t)+βa(t)+γc(t)+δf(t)

r _d (t)=αd(t)+βe(t)+γc(t)+δf(t)

And 5.1 to 1-channel downstream mix:

The values commonly used for the constants α, β, γ, and δ are: α = 1, β = γ = , and δ=0.

The IID parameter is defined as the energy ratio of the channels of two randomly selected channels or weighted groups. If there is the energy of the channel outlined above for the 5.1 channel configuration, then several sets of IID parameters can be defined.

7 shows a general view of a downlink mixer 700, using the above equation in order to calculate a best two mono or stereo channels m l _d and r _d. Typically, this downstream mixer uses some downstream mix information. In a preferred embodiment of a linear downmix, the downmix information includes weighting factors α, β, γ, and δ. It is known in the art that more or less constant or non-constant weighting factors can be used.

In an ITU recommended downmix, the alpha system is set to 1, the beta and gamma systems are set equal to the square root of 0.5, and the delta system is set to zero. Generally, the factor a can vary between 1.5 and 0.5. Further, the factors β and γ are different from each other and vary between 0 and 1. This low frequency enhanced channel f(t) has the same fact. The factor δ of this channel can vary between 0 and 1. Furthermore, the factors of the left-downmix and the right-downmix do not have to be equal to each other. This becomes clearer when considering a non-automatic downmix performed by a sound engineer, for example. Further directing the sound engineering to implement a creative downmix rather than a mix of sounds under the control of any mathematical law. Instead, the sound engineer is dominated by his own creative feelings. When a certain parameter group records this "creative" downstream mix, the "creative" downstream mix will be used in accordance with the invention by the invention of the upstream mixer shown in Fig. 8, which is not only by the parameters It is dominated and is also dominated by additional information about the downstream mix architecture.

When a linear downmix is implemented as in Figure 7, the weighting parameters are the best information to be used by the upstream mixer on the downstream mix architecture. However, when presenting other information used in the downlink mixing architecture, an upstream mixer can also use this other information as information for the downstream mixing architecture. Such other information may also be, for example, certain matrix elements or certain factors or functions within a matrix element of an upstream mix-matrix (eg, as shown in FIG. 11).

Suppose there is a 5.1-channel configuration as outlined in Figure 1 and how the other channel configurations are related to the 5.1-channel configuration: for 3-channel situations where surround channels are not available, Marks get B, C and D. For a 4-channel configuration, B, C, and D are available, however a combination of A and E to represent the single surround channel or the subsequent channels in this context is also available.

The present invention uses IID parameters applied to all of these channels, i.e., the 4-channel sub-group of the 5.1 channel configuration has a corresponding sub-group within the IID parameter set describing the 5.1 channel. The following IID parameter group resolves this issue:

It is apparent that the r ₁ parameter corresponds to the energy ratio between the left downmix channel and the right down mix channel. The r ₂ parameter corresponds to the energy ratio between the center channel and the left and right front channels. The r ₃ parameter corresponds to the energy ratio between the three front channels and the two surround channels. The r ₄ parameter corresponds to the energy ratio between the two surround channels. The r ₅ parameter corresponds to the energy ratio between the LFE channel and all other channels.

In Figure 4, the energy ratios described above are described. The different output channels are denoted by 101 to 105 and are identical to those shown in Fig. 1 and thus will not be described in detail herein. The speaker assembly is divided into left and right halves, wherein the center channel 103 is part of two halves. The energy ratio between the left half plane and the right half plane is exactly the r ₁ parameter. This is indicated by the solid line below r _{1 in} Figure 4. Furthermore, the energy distribution between the center channel 103 and the left front 102 and right front 103 channels is represented by r ₂ . Finally, installation (102, 103, and 104) in the entire front channel and the rear channel energy (101 and 105) between the distribution system to be described by the parameters r ₃ in FIG. 4 of the arrows.

Suppose there is the above parameterization and the energy of the single downlink mixing channel of the transmission:

The energy of the reconstructed channels can be expressed as:

Thus, the energy of the M signal can be distributed to the reconstructed channels, resulting in the reconstructed channels having the same energy as the original channels.

The preferred upstream mixing architecture described above is depicted in FIG. From the equations of F, A, E, C, B, and D, it is clear that the information used by the upstream mixer for the downlink mixing architecture is the weighting factors α, β, γ, and δ, which are weighted. Or the unweighted channels are added together or deducted from each other to obtain a certain number of lower line mixing channels, the weighting factors are used to weight the original channels, wherein the number of downstream mixing channels is less than the original channel number. Therefore, it is clear from Fig. 8 that the energy of the reconstructed channels according to the present invention is determined not only by the balance parameters transmitted from an encoding side to a decoding side, but also by the downstream mixing factors α, β, γ, and δ. .

When considering Fig. 8, it becomes clear that in order to calculate the left and right energies B and D, the calculated vocal energy F, A, E, and C can be used in the equation. However, it is not necessary to include a continuous upstream mixing architecture. Instead, in order to obtain a fully parallel upstream mixing architecture implemented using, for example, an upstream mixing matrix (with certain upstream mixing matrix elements), the equations of A, C, E, and F are inserted into the equations of B and D. Therefore, it becomes clear that the reconstructed channel energy is determined only by the balance parameters, the downmix channel, and the information of the downmix architecture (eg, the downmix factors).

As will be apparent from the following, assuming the above IID parameters, it is apparent that the problem of defining an IID parameter of a parameter set (available for several channel configurations) has been solved. Observing the three-channel configuration (that is, reconstructing three front channels from an available channel) as an example, it is obvious that since the A, E, and F channels do not exist, r ₃ , The r ₄ and r ₅ parameters are not significant. It can also be clearly seen that since the parameter r ₁ describes the energy ratio between the left and right front channels and the parameter r ₂ describes the energy ratio between the center channel and the left and right front channels, the parameters r ₁ and r ₂ are sufficient. The three channels are reconstructed from the next line of mixing.

In a more general case, it can be easily seen that the above IID parameters (r ₁ ... r ₅ ) are applied to all subgroups for reconstructing n channels from m channels, where m < n 6. Looking at Figure 4, you can say:

- for a system that reconstructs 2 channels from 1 channel, obtaining sufficient information from the r ₁ parameter to maintain the correct energy ratio between the channels;

- for a system that reconstructs 3 channels from 1 channel, obtaining sufficient information from the r ₁ and r ₂ parameters to maintain the correct energy ratio between the channels;

- for a system that reconstructs 4 channels from 1 channel, obtaining sufficient information from the r ₁ , r ₂ and r ₃ parameters to maintain the correct energy ratio between the channels;

- For a system that reconstructs 5 channels from 1 channel, sufficient information is obtained from the r ₁ , r ₂ , r ₃ and r ₄ parameters to maintain the correct energy ratio between the channels;

- For a system that reconstructs 5.1 channels from 1 channel, sufficient information is obtained from the r ₁ , r ₂ , r ₃ , r ₄ and r ₅ parameters to maintain the correct energy ratio between the channels;

- For a system that reconstructs 5.1 channels from 2 channels, sufficient information is obtained from the r ₂ , r ₃ , r ₄ and r ₅ parameters to maintain the correct energy ratio between the channels.

The above adjustable capability features can be described by the list in Figure 10b. The tunable bit stream described in Figure 10a and described later may also be applied to the list in Figure 10b in order to obtain a finer tunable capability than those described in Figure 10a.

The advantage of the invention is in particular that the left and right channels can be easily reconstructed from a single balancing parameter r ₁ without knowing or taking any other balancing parameters. For this purpose, in the equations of B and D in Fig. 8, the channels A, C, F, and E are simply set to zero.

In another case, when only the balance parameter r ₂ is considered, the reconstructed channels are the sum of the center channel and the low frequency channel (the channel is not set to zero) and the left and right The sum of the channels. Therefore, the center channel and the tone signal can be reconstructed using only a single parameter. This feature is useful for a simple 3-channel representation in which the left and right signals are obtained, for example, by summing from left and right, and wherein the central and correct decisions are made by the balancing parameter r ₂ The energy between the left and right.

In this context, the equalization parameters r ₁ or r ₂ are located in a lower adjustment layer.

As for the second item in the list of Figure 10b, it shows how to replace all five balance parameters with only two balance parameters to produce the sum of the three channels B, D and C and F, which is lower than the lower one. The parameter r ₁ or r ₂ in the layer is adjusted, one of these parameters r ₁ and r ₂ may already be in a higher adjustment layer.

When considering the equation in Fig. 8, it becomes clear that in order to calculate C, the untaken parameter r ₅ and the other non-taken parameter r _{3 are} set to zero. In another case, the unused channels A, E, and F are also set to zero so that the three channels B, D and the combination of the center channel and the low frequency enhanced channel F can be calculated. .

When a 4-channel is indicated for the upstream mix, it is sufficient to only extract r ₁ , r ₂ and r ₃ from the parameter data stream. In this context, r ₃ can be in a next higher adjustment layer than the parameter r ₁ or r ₂ . Since the third balancing parameter r ₃ has been obtained from the combination of the front channel and the rear channels as described later with respect to FIG. 6, the 4-channel configuration is particularly suitable for the present invention. Overbalanced parameter representation. This is based on the fact that the parameter r ₃ is a pre-post balance parameter obtained from the pair of channels, the pair of channels having the combination of the rear channels A and E (as the first channel) and It has a combination of left channel B, right channel E and center channel C (as the front channel).

Thus, as is the case in a single parametric channel setup, the combined channel energy of the two surround channels can be automatically obtained without any further separation calculations and subsequent combinations.

When five channels must be reconstructed from a single channel, another balancing parameter r _{4 is required} . This parameter r ₄ can again be located in the next higher adjustment layer.

Each balance parameter is required when a 5.1 reconstruction must be performed. Therefore, the next higher adjustment layer (including the next balance parameter r ₅ ) must be transmitted to and estimated by the receiver.

However, using the same method of augmenting the IID parameter depending on the number of expansions of the channel, the above IID parameters can be extended to cover a channel configuration having a larger number of channels than the 5.1 configuration. Accordingly, the invention is not limited to the examples outlined above.

Now observe that the channel configuration is a 5.1 channel configuration, which is one of the most common use cases. Again, assume that the 5.1 channel is reconstructed from two channels. For this case, the parameters r ₃ and r ₄ can be replaced by the following formula to define a different set of parameters:

The parameters q ₃ and q ₄ represent the energy ratio between the front and rear left channels and the energy ratio between the front and rear right channels. Imagine a few other parameterizations.

In Figure 5, the parameterization of the modification can be seen. Instead of having a parameter for summarizing the energy distribution between the front and rear channels (as outlined by r ₃ in FIG. 4) and one for describing the energy distribution between the left surround channel and the right surround channel (As outlined in r ₄ of FIG. ₄ ), the parameters q ₃ and q _{4 are used} to describe the energy ratio between the left front 102 and the left surround 101 and the right front channel 104 and the right surround channel 105. The energy ratio between the two.

The teachings of the present invention may use several sets of parameters to represent the multi-channel signals. An additional feature of the present invention is that different parameterizations can be selected depending on the type of quantization of the parameters used.

Taking a system using parameterized coarse quantization as an example, due to the high bit rate limitation, a parameterization that does not expand the error in the upstream mixing program should be used.

Observe two representations of the above reconstructed energy in a system for reconstructing 5.1 channels from one channel:

It is apparent that due to the relatively small quantization effect of the M, A, C and F parameters, such subtraction produces a large change in the B and D energy.

In accordance with the present invention, a different parameterization that is not sensitive to the quantification of the parameters should be used. Therefore, if coarse quantization is used, the r ₁ parameters defined above are:

Can be replaced by an alternative definition according to the following formula:

This produces an equation for the reconstruction energy according to the following formula:

The equations for the reconstruction energies of A, E, C, and F remain the same as described above. It is apparent that this parameter represents an optimal state system from a quantitative point of view.

In Fig. 6, the energy ratios described above are described. The different output channels are denoted by 101 to 105 and are identical to Figure 1 and therefore will not be described in further detail herein. The speaker assembly is divided into a front portion and a rear portion. The energy distribution between by the arrow in FIG. 6 indicated by the r ₃ parameter to describe the whole of the front mounting channel (102, 103, and 104) with such rear channel (101 and 105).

Another important distinguishing feature of the present invention is that when the parameterization is observed

From a quantitative point of view, it is not only a better state system. The above parameterization also has the advantage that parameters for reconstructing the three front channels can be obtained without any effect on the surround channels. The relationship between the center channel and all other channels can be described as a parameter r ₂ system. However, this would have the disadvantage of the following example: the surround channels will be included in the estimates of the parameters described in the front channels.

It is to be noted that the parameterization described in the present invention can also be applied to the correlation or coherence measurement between channels, and it is apparent that the success of reconstructing the front channels by including the rear channel pairs in the calculation of r ₂ is known. There are significant negative effects.

It can be seen as an example of the same signal in all front channels and the absence of relevant signals in the back channels. This is not uncommon, assuming that the back channels are often used to reconstruct the surrounding environment of the original sound.

If the description of the center channel is for all other channels, the degree of correlation between the center and all other channels will be quite low because the back channels are completely uncorrelated. The same fact is used for a parameter used to estimate the correlation between the front left/right channel and the rear left/right channel.

Therefore, we have reached a parameterization that correctly reconstructs these energies, however this parameterization does not include all of the same (ie, very relevant) information about the front channel. The parameterization does include decoupling the left and right front channels to the back channels and also correlating the center channels to the back channels. However, the fact that all front channels are the same cannot be inferred from this parametric.

Since the back channels are not included in the estimate of the parameters used on the decoder side to reconstruct the front channels, this can be overcome by using the following formula taught by the present invention:

The energy distribution between the center channel 103 and the left front 102 and right front 103 channels is represented by r _{2 in} accordance with the present invention. The energy distribution between the left surround channel 101 and the right surround channel 105 is described by r ₄ . Finally, r ₁ is provided by the front left channel 102 and the distribution of energy between the right front channel 104. It is apparent that all parameters are the same as described in FIG. 4 except that r ₁ corresponds here to the energy distribution between the left front speaker and the right front speaker (as opposed to the entire left and the entire right side). Based on the integrity, the parameter r _{5 is} also provided to summarize the energy distribution between the center channel 103 and the LFE channel 106.

Figure 6 shows an overview of a preferred parameterized embodiment of the present invention. The first balance parameter r ₁ (represented by the solid line) constitutes a front-left/front-right balance parameter. The second balance parameter r ₂ is a central left-right balance parameter. The third balance parameter r ₃ constitutes a pre/post balance parameter. The fourth balance parameter r ₄ constitutes a back-left/back-right balance parameter. Finally, the fifth parameter r ₅ constitutes a central/LFE balancing parameter.

Figure 4 shows a related situation. The first balance parameter r ₁ (described by the solid line in FIG. 4 in a downmix left/right balance) may be defined by a channel B and D (lower channel pair) The original front-left/pre-right balance parameters are replaced. This is described by the broken line r _{1 in} FIG. 4 and corresponds to the solid line r _{1 in} FIGS. 5 and 6.

In the case of a two-channel, the parameters r ₃ and r ₄ (i.e., the front/rear balance parameters and the back-left/right balance parameters) are replaced by two one-sided front/rear parameters. The first one-sided front/rear parameter q ₃ can also be regarded as the first balance parameter, wherein the first balance parameter is obtained from the channel pair formed by the left surround channel A and the left channel B . The second one-sided front/left balance parameter is the parameter q ₄ , which can be regarded as the second parameter, and the second parameter is based on the second sound formed by the right channel D and the right surround channel E Right. Furthermore, the two channel pairs are not related to each other. The central/left-right balance parameter r ₂ also has the same fact, the central/left-right balance parameter r ₂ has a center channel C as a first channel and the left and right channels B and D The sum is used as a second channel.

Another parameterization is defined in accordance with the present invention, which is quite suitable for coarse quantization for a system that reconstructs 5.1 channels from one or two channels.

As for one channel to 5.1 channel:

and

As for the case of two channels to 5.1 channels:

and

It is apparent that the above parameterization includes more parameters than are required by the strict theoretical point of view to correctly redistribute the energy of the transmitted signals to the reconstructed signals. However, the sensitivity of this parameterization to quantization errors is very slow.

The above mentioned reference sets for a 2-channel installation use several reference channels. However, compared to the parameter configuration in Figure 6, the parameter set in Figure 7 is only used as the reference channel based on the downmix channel instead of the original channel. The equalization parameters q ₁ , q ₃ and q ₄ are obtained from completely different pairs of channels.

Although several embodiments of the invention have been described in which the channel pairs used to obtain the balance parameters include only the original channels (Figs. 4, 5, and 6) or include the original channels and the downmix channels ( 4 and 5) or only the downlink mixing channel as the reference channel represented at the bottom of FIG. 7, but preferably included in the surround encoder 206 of FIG. The parameter generator is operative to use only a combination of the original channel or the original channel rather than a combination of a base channel or a base channel of the channel in the pair of channels, wherein the equalization parameters are based on the Channel pair. This is due to the inability to fully guarantee that the single base channel or the two stereo base channels will not undergo an energy change during transmission from a surround encoder to a surround decoder. The downlink mixer channel or the single downmix channel can be caused by an audio encoder 205 (Fig. 2) or an audio decoder 302 (Fig. 3) operating in a low-bit rate state. The energy changes. This condition may result in manipulation of the energy of the single downmix channel or the stereo downmix channels, which may be different or even frequency selective between the left and right stereo downmix channels Or time selective.

In order to completely and safely oppose this energy change, an additional level parameter is transmitted in accordance with the present invention for each region and frequency band of each downstream mixing channel. Therefore, when the equalization parameters are based on the original signal rather than the downmix signal, a single correction factor is sufficient for each band because any energy correction will not affect the balance between the original channels. Even when no extra level parameters are transmitted, any downmix channel energy changes will not cause distortion localization of the source in the audio image, but will only result in a general volume change, which will not be like a normal volume change. The migration of sound sources caused by changing the balance state is as annoying.

It is important to note that care must be taken so that the energy M (of the downstream mixing channels) is the sum of the energies B, D, A, E, C and F outlined above. This is not always the case due to the phase dependence between the different channels being downmixed to one channel. The energy correction factor can be transmitted as an additional parameter r _M , and thus the line mix signal received on the decoder side is defined as:

In Figure 9, an application of the additional parameter r _M in accordance with the present invention is outlined. The downlink mix signal is modified by the additional parameter r _M in 901 before the downlink mix signal is transmitted to the upstream mix module 701-705. These are the same as those described in Figure 7 and will not be described in further detail herein. It will be apparent to those skilled in the art that the parameters r _{M of the} above mono downmixing paradigm can be extended to one parameter per downstream mix and thus are not limited to a single downmix channel.

Figure 9a depicts an inventive level parameter calculator 900, while Figure 9b shows an inventive level corrector 902. Figure 9a shows the situation on the encoder side, and Figure 9b depicts the corresponding situation on the decoder side. The level parameter or "extra" parameter r _M is used to provide a correction factor for a certain energy ratio. The following exemplary scenarios are assumed to be explained. For a certain original multi-channel signal, on one hand, there is a "mainstream downmix" and on the other hand, a "parameter downmix". The primary downmix has been generated by a sound engineer in a studio based on, for example, a subjective quality impression. In addition, an audio storage medium also includes the parameter downlink mix, which is implemented by, for example, the surround encoder 203 of FIG. The parameter downmix includes a base channel or two base channels that use the balance parameter set of the original multichannel signal or any other parameter representation to form the basis for the multichannel reconstruction.

For example, it may be the case that the broadcaster wishes to not transmit the parameter downmix, but it is desirable to transfer the primary downmix from a transmitter to the receiver. In addition, in order to promote the main downmix to a multi-channel representation, the announcer also transmits a parameter representation of the original multi-channel signal. Since (in a frequency band and in a block) energy can (or will typically) vary between the primary downmix and the parameter downmix, a relative level parameter r is generated in block 900. _M and pass it to the receiver as an additional parameter. The level parameter is obtained from the main downmix and the parameter downmix and preferably the ratio of the energy in one of the main downmix and the parameter downmix and a band.

Typically, the level parameter is calculated to be the ratio of the sum of the energy of the original channels (E _orig ) to the energy of the (equal) downmix channel, wherein the (equal) downmix channel can Is the parameter Downmix (E _PD ) or the Main Downmix (E _MD ) or any other downstream mix. Typically, the energy of a particular downstream mix signal transmitted from an encoder to a decoder is used.

Figure 9b depicts the implementation of this level parameter using one of the decoder sides. The level parameter and the downmix signal are input to the level corrector block 902. The level corrector corrects the single base channel or the base channels according to the level parameter. Since the additional parameter r _M is a relative value, the relative value is manipulated by the energy of the corresponding basic channel.

Although the figures 9a and 9b show the case where a level correction is applied to the pair of downmix channels or the downmix channels, the level parameters can also be integrated into the upmix matrix. For this purpose, each occurrence of m in the equation of Fig. 8 is replaced by "r _M M".

When the 5.1 channel is reconstructed from 2 channels, the following description can be observed.

If the invention with codecs 205 and 302 as outlined in Figures 2 and 3 is used, some more considerations are required. Observe the IID parameter defined earlier, where r ₁ is defined according to the following formula:

Since the system reconstructs 5.1 channels from 2 channels, assuming that the two transmission channels are stereo downmixes of the surround channels, this parameter is implicitly available on the decoder side.

However, an audio codec operating at a one-bit rate limit can modify the spectral distribution such that the measured L and R energies on the decoder side are different from the values on the encoder side. According to the present invention, the effect of the energy distribution on the reconstructed channel disappears by transmitting the following parameters for the case of reconstructing 5.1 channels from two channels:

If a means of signaling is provided, the decoder can encode the current signal segment using different parameter sets and select an IID parameter to provide the lowest burden on the particular signal segment to be processed. The energy levels between the right front and rear channels may be similar, and the energy levels between the front and rear left channels may be similar, however the levels in the right front and back channels are significantly different. . Assuming that there is delta coding of the parameters and subsequent entropy coding, it is more efficient to use the parameters q ₃ and q ₄ instead of the r ₃ and r ₄ systems. For another signal segment with different characteristics, a different set of parameters can provide a lower bit rate burden. The present invention allows for free switching between different parameter representations in order to minimize the bit rate burden of the currently encoded signal segment, wherein the characteristics of the signal segment are known. The ability to switch between different parameterizations of the IID parameters to obtain the lowest possible bit rate burden and to provide means of signaling to indicate what parameterization is currently used is an essential feature of the present invention.

Furthermore, the difference encoding of the parameters can be done in the frequency direction or in the time direction, and the difference encoding between the different parameters can be completed. In accordance with the present invention, assuming that a means of signaling is provided to indicate the particular delta encoding used, a parameter can be differentially encoded with respect to any other parameter.

An important feature of any coding architecture is the ability to implement tunable coding. This means that the encoded bit stream can be split into several different layers. The core layer can be decoded by itself, and the higher layer can be decoded to enhance the decoded core layer signal. The number of available layers may vary for different situations, but as long as the core layer is available, the decoder can produce output samples. The parameterization of the multi-channel coding outlined above using the r ₁ to r ₅ parameters is itself well suited for tunable coding. Therefore, for example, the data of the two surround channels (A and E) can be stored in an enhancement layer (that is, the parameters r ₃ and r ₄ and the parameters corresponding to the front channels in a core layer) (indicated by parameters r ₁ and r ₂ )).

In Fig. 10, an implementation of an adjustable bit stream in accordance with the present invention is outlined. The bit stream layer is described by 1001 and 1002, wherein 1001 is the core layer, which holds the waveform encoded downmix signal and holds the front channel (102, 103 and 104) for reconstruction. Parameters r ₁ and r ₂ . The enhancement layer described in 1002 holds parameters for reconstructing the back channels (101 and 105).

Another important aspect of the present invention is the use of a decorrelator in a multi-channel configuration. The idea of the use of a correlator has been detailed in the PCT/SE02/01372 patent document for one or two channel cases. However, when this theory is extended to more than two channels, several problems to be solved by the present invention arise.

Basic Mathematical Display: In order to complete M mutually decorrelated signals from N signals, MN decorrelators are required, with all of the different decorrelators used to generate a plurality of mutually orthogonal output signals from a common input signal. Suppose an input x(t) produces an output y(t) and And almost the interaction correlation E [ xy ^* ] disappears, then a decorrelator is usually an all-pass or almost all-pass filter. Additional perceptual criteria can be used to obtain a good decorrelator design. Some examples of design methods can minimize the comb filter characteristics and minimize one of the transient signals when adding the original signal to the decorrelated signal. The effect of an impulse response that is too long. Some conventional art decorators use an artificial mirror to decorrelate. Conventional techniques can also achieve higher echo densities and thus longer diffusions by, for example, modifying the phase of complex sub-band samples to include fractional delays.

The present invention proposes a method for modifying a mirror-based decorrelator to achieve a plurality of decorrelators that can generate a plurality of mutually decorrelated output signals from a common input signal. If the outputs y ₁ (t) and y ₂ (t) of the two decorrelators have an alternating correlation that disappears or almost disappears (assuming the same input), then the two decorrelators are decorrelated to each other. Assuming that the input is static white noise, then on E [ The impulses h ₁ and h ₂ must be orthogonal in the perception of disappearance or almost disappearance. The pairwise de-correlation decorrelator of a complex array can be constructed in several ways. One effective way to implement this modification is to change the phase rotation factor q (which is part of the fractional delay).

The particular phase rotation factor of the present invention may be part of the delay line in the all-pass filter or just a total fractional delay. In this latter case, the method is not limited to all-pass or mirror filters, but can also be applied to, for example, a simple delay including a fractional delay. One of the all-pass filter connections in the decorrelator can be described in a Z-domain as:

Where q is the complex phase rotation factor (| q |=1), m is the delay line length in the sample, and the a-line filter coefficient. For stability reasons, the size of the filter coefficient must be limited to | a |<1. However, by using the alternative filter coefficients a' = -a to define a new mirror that has the same reflection delay characteristics, however, has an output that is significantly uncorrelated with the output of the unmodified mirror. Furthermore, the modification of the phase rotation factor q can be accomplished by, for example, adding a fixed phase offset q'=qe ^jC . This constant C can be used as a fixed phase offset or can be adjusted in such a way that for all frequency bands to which the constant C is applied, the constant C will correspond to a fixed time offset. The phase offset constant C can also be a random value that is different for all frequency bands.

According to the present invention, by applying an upstream mixing matrix H having an n×(m+p) size to a row vector signal having a size of (m+p)×1, n generations are generated from m channels. Channel.

Where m is the m downmixed and encoded signals, and the p signals in s are mutually decorrelated and decoupled from all signals in m. These decorrelated signals are generated by the signal in m by the decorrelator. Then, n reconstruction signals a', b', ... are included in the row vector.

x'=Hy.

The above is described by Figure 11, wherein the decorrelated signals are generated by the decorrelators 1102, 1103, and 1104. The upstream mixing matrix H is provided by 1101 to operate on the vector y to provide the output signal x'.

Let R = E[xx ^* ] be the correlation matrix of the original signal vector, assuming R' = E[x'x' ^* ] is the correlation matrix of the reconstructed signal. Here and in the following, for a matrix having a vector X of complex terms, X ^* represents a complex conjugate transpose of the adjoint matrix --X.

The diagonal of R contains the energy values A, B, C... and can be decoded into a total energy level by the energy rating defined above. Since R ^* = R, there are only n(n-1)/2 different non-diagonal cross-correlation values, which contain information that will be completely or partially reconstructed by adjusting the upstream mix matrix H. The reconstruction of the complete correlation structure corresponds to the case R'=R. The reconstruction of the correct energy level corresponds only to the case where R' and R are equal on the diagonal.

In the case of changing from m = 1 channel to n channel, the reconstruction of the complete correlation structure is achieved by using p = n - 1 mutual decorrelation decorrelator (an upstream mixing matrix H), wherein the upmix The tone matrix H satisfies the following conditions:

Where M is the energy of the single transmitted signal. Because R is a positive semi-definite matrix, it is well known to be an answer now. Again, n(n-1)/2 degrees of freedom are reserved for the design of H, which is used in the present invention to obtain additional desirable characteristics of the upstream mixing matrix. A central design criterion for H should be responsive to the transmission-related data.

One of the conventional ways of parameterizing the upstream mixing matrix is H=UDV, where U and V are orthogonal matrices and D is a pair of angular matrices. The square of the absolute value of D can be selected to be equal to the eigenvalue of R/M. Deleting V and selecting the eigenvalues to apply the maximum value to the first coordinate will minimize the total energy of the decorrelated signal in the output. In the real case, the orthogonal matrix U is parameterized by an n(n-1)/2 rotation angle. Transmitting the relevant data in the form of those angles and the n diagonal values of D will immediately provide the desired smooth compliance of H. However, because the energy data must be transformed into eigenvalues, this approach sacrifices the ability to adjust.

The second method taught by the present invention defines a normalized correlation matrix R ₀ by R = GR ₀ G to separate the energy portion in R from the correlation portion, wherein the G system has a diagonal term equal to R The diagonal value of the square root (ie, , The diagonal matrix of ...), R ₀ has the same diagonal value on the diagonal. It is assumed that H0 is an orthogonal up-mixing matrix that defines a better normalized upstream mix in the case of a completely unrelated signal of equal energy. An example of this preferred upstream mixing matrix is:

Then, with H = GSH ₀ / Defined upstream mixing, a solution wherein the matrix S SS ^* = R _0. This dependency on the selected answers normalized cross-correlation values in R ₀ is in the continuous matrix S is equal to in the case of the R ₀ = I.

Segmenting the n channels into groups of fewer channels is a convenient way to reconstruct a portion of the cross-correlation structure. According to the present invention, for the case of reconstructing 5.1 channels from 1 channel, a particularly advantageous grouping is {a, e}, {c}, {b, d}, {f}, where no decorrelation is applied to the The equal groups {c} and {f}, and the groups {a, e} and {b, d} are generated by the same downmixing/de-correlated pair of upstream mixes. For both subsystems, the preferred normalized upstream mix in the completely unrelated case is selected as:

Therefore, only two of the 15 cross-correlation totals will be transmitted and reconstructed, that is, the interaction between the channels {a, e} and {b, d}. Among the terms used above, this is an example of design for the case of n=6, m=1, and p=1. The upstream mixing matrix H is 6 x 2 in size and the two terms corresponding to the outputs c' and f' in the second row on the third and sixth columns are zero.

A third method for incorporating a decorrelated signal as taught by the present invention is a simpler view: each output channel has a different decorrelator to cause decorrelated signals s _a , s _b .... Then make the reconstruction signals into:

and many more.

The parameters φ _a , φ _b ... control the number of decorrelated signals presented in the output channels a', b'.... This related information is transmitted in the form of these angles. It can be easily calculated that the normalized interactive phase relationship between, for example, the channels a' and b' is equal to the product cosφ _a cosφ _b . When the number of pairwise interaction correlations is n(n-1)/2 and there are n decorrelators, if n>3, it is usually impossible to match a specific correlation structure in this way, but the advantage is that it is very simple and A stable decoding method and direct control of the number of generated decorrelation signals presented in each output channel. This enables the mixture of decorrelated signals to be based on a perceptual criterion incorporating energy level differences such as channel pairs.

For the case of reconstructing the n channel from m>1 channel, the correlation matrix R _y =E[yy ^* ] is no longer assumed to be a diagonal matrix, and R'=HR _y H ^* must be considered for the target R Match. Because Ry has a block matrix structure

So a simplification is produced, where R _m = E[mm ^* ] and R _s = E[ss ^* ]. Furthermore, it is assumed that the decorrelator is mutually de-correlated, and the matrix R _s is a diagonal matrix. Note that this also affects the upstream mix design with respect to the reconstruction of the correct energy. The solution is to calculate or transmit from the encoder information about the correlation structure R _m of the downstream mix signals in the decoder.

For the case of reconstructing 5.1 channels from 2 channels, the preferred method of upstream mixing is:

Where s ₁ can be obtained from the decorrelation of m ₁ = l _d and s ₂ can be obtained from the decorrelation of m ₂ = r _d .

Here, the groups {a, b} and {d, e} are regarded as separate 1→2 channel systems that have been considered to be pairwise interactive. For channels c and f, adjust the weighting so that

E | h ₃₁ m ₁ + h ₃₂ m ₂ | ² = C ,

E | h ₆₁ m ₁ + h ₆₂ m ₂ | ² = F .

The present invention can be implemented in hardware chips and DSPs for a variety of systems using arbitrary codecs for the storage or transmission of analog or digital signals. Figures 2 and 3 show possible implementations of the invention. In this example, a system for operating six input signals (a 5.1 channel configuration) is shown. In the second diagram showing the encoder side, the analog input signals, such as separate channels, are converted into digital signals 201 and analyzed 202 using a filter bank for each channel. The output of the filter bank is fed to the surround encoder 203, which includes a parameter generator that performs a line mix to produce one or two channels encoded by the audio encoder 205. Furthermore, in accordance with the present invention, surround parameters such as IID and ICC parameters are retrieved, and a time frequency grid for summarizing the data and which parameterized control data 204 are used in accordance with the present invention are retrieved. As taught by the present invention, the capture parameters 206 are encoded to switch between different parameterizations or to configure the parameters in an adjustable manner. The surround parameters 207, control signals, and numbered downmix signal 208 multiplex processing 209 are a series of bitstreams.

In Figure 3, a typical decoder implementation (i.e., a means for generating multi-channel reconstruction) is shown. Here, it is assumed that the audio decoder outputs a signal in a frequency domain representation, for example, the output of an MPEG-4 high efficiency AAC decoder in front of the QMF synthesis filter bank. Performing a demultiplexing process 301 on the serial bit stream and feeding the encoded surround data into the surround data decoder 303 and feeding the downlink mixed code channels into the audio decoder 302 (in this example, MPEG-4 high efficiency AAC decoder). The surround data decoder decodes the surround data and feeds it into the surround decoder 305. The surround decoder 305 includes an upstream mixer, according to the decoded downlink mix channel and the surround data and the control signals. To reconstruct 6 channels. The frequency domain output of the surround decoder is synthesized 306 to become a time domain signal, which is then converted to an analog signal by the DAC 307.

Although the invention has been described primarily with respect to the generation and use of balance parameters, it is emphasized here that the same grouping of channel pairs for obtaining balanced parameters is preferably used to calculate inter-channel coherence parameters or both. The "width" parameter between the pairs. In addition, an inter-channel time difference or a "phase signal" can also be obtained using the same pair of channels used for the calculation of the balance parameter. On the receiver side, these parameters can be used in addition to or as an alternative to the equalization parameters to produce a multi-channel reconstruction. In another case, the inter-channel coherence parameters or even the inter-channel time differences may be used in addition to other inter-channel level differences as determined by other reference channels. However, in view of the adjustable capability features of the present invention as described in Figures 10a and 10b, it is preferred to use the same pair of channels for all parameters so that each adjustment layer is included in an adjustable bit stream. All parameters of the output channel of the subgroup are reconstructed, wherein the output channels of the subgroup are generated by the individual adjustment layers outlined in the penultimate row of the list in Figure 10b. The present invention is useful in calculating only the coherence parameters or time difference parameters between individual channel pairs and transmitting them to a decoder. In this case, when a multi-channel reconstruction is implemented, the level parameters are already present at the decoder for use.

The process of the invention may be carried out in a hard or soft manner in accordance with certain embodiments of the method of the invention. It can be implemented using a digital storage medium, particularly a magnetic disk or optical disk storing electronically readable control signals, which cooperate with a programmable computer system to carry out the method of the present invention. Accordingly, the present invention is generally a computer program product having a code stored in a mechanically readable carrier, the program being operative to carry out the method of the present invention when the computer program product is executed on a computer. Thus, in other words, the method of the present invention is a computer program having a program code for performing at least one of the methods of the present invention when the computer program is executed on a computer.

101. . . Left surround channel

102. . . Left front channel

103. . . Central channel

104. . . Right front channel

105. . . Right surround channel

106. . . LEF channel

201. . . ADC

202. . . Analysis filter bank

203. . . Surround encoder

204. . . control signal

205. . . Audio encoder

206. . . Surround data encoder

207. . . Surround parameter

208. . . Numbered downmix signal

209. . . Multiplexer

301. . . Demultiplexer

302. . . Audio decoder

303. . . Surround data decoder

304. . . control signal

305. . . Surround decoder

306. . . Synthesis filter bank

307. . . DAC

700. . . Downstream mixer

900. . . Level parameter calculator

902. . . Level corrector

1001. . . Bit stream layer

1002. . . Bit stream layer

1101. . . Decomposer

1103. . . Decomposer

1104. . . Decomposer

l _d , r _d . . . Best stereo channel

m,l,n. . . Mono

r ₁ . . . parameter

r ₂ . . . parameter

r ₃ . . . parameter

r ₄ . . . parameter

r ₅ . . . parameter

r _M . . . Additional parameters

α. . . Weighting factor

β. . . Weighting factor

γ. . . Weighting factor

δ. . . Weighting factor

A. . . Channel

B. . . Channel

C. . . Channel

D. . . Channel

E. . . Channel

E _MD . . . Main downmix

E _orig . . . Original channel energy

E _PD . . . Parametric downmix

F. . . Rebuilding the energy of the channel

Figure 1 depicts the academic terms used in one of the 5.1 channel configurations of the present invention;

Figure 2 depicts a suitable encoder implementation of one preferred embodiment of the present invention;

Figure 3 depicts a suitable decoder implementation of one preferred embodiment of the present invention;

Figure 4 depicts a preferred parameterization of a multi-channel signal in accordance with the present invention;

Figure 5 depicts a preferred parameterization of a multi-channel signal in accordance with the present invention;

Figure 6 depicts a preferred parameterization of a multi-channel signal in accordance with the present invention;

Figure 7 depicts a schematic setup for generating a single base channel or two basic channel sub-mixing architectures;

Figure 8 depicts a schematic representation of an upstream mixing architecture that balances parameters and information about the downstream mixing architecture in accordance with the present invention;

Figure 9a is a diagrammatic description of the decision of the level parameter on the encoder side in accordance with the present invention;

Figure 9b is a diagrammatic description of the use of level parameters on the decoder side in accordance with the present invention;

Figure 10a depicts an adjustable bit stream having different portions of the multi-channel parameterization in different layers of the bitstream;

Figure 10b depicts an adjustable capability table indicating which balance parameters are used to construct which channels and which balance parameters and channels are not used and calculated; and Figure 11 depicts the application of the upstream mix matrix in accordance with the present invention.

902. . . Level corrector

m,l,n. . . Mono

r _M . . . Additional parameters

Claims (8)

A device for generating a level parameter in a parameter representation of a multi-channel signal, the multi-channel signal having a plurality of original channels, the parameter representation comprising a parameter group, the parameter group being at least The multi-channel reconstruction is allowed when the mixing channel is used together. The device comprises: a level parameter calculator (900) for calculating a level parameter (r _M ), the level parameter is in a main downmix a level difference between a tone and a parameter downmix, wherein the parameter indicates a downmix according to the parameter; an output interface for generating an output data, the output data including the level parameter and the parameter group or the a flat parameter and the at least one lower mixing channel, and a parameter generator formed to generate the parameter set, wherein a left/right balance parameter is used as a first balance parameter, a central balance The parameter is used as a second balance parameter, a front/back balance parameter as a third balance parameter, a rear-left/right balance parameter as a fourth balance parameter, and a low frequency enhancement balance parameter as a fifth Balance parameters. The apparatus of claim 1, wherein the parameter comprises a parameter group for each of a plurality of frequency bands of the at least one of the next mixing channels, and wherein the parameter calculator (900) operates A level parameter is calculated for each of the bands. The device of claim 1, wherein the parameter representation comprises a parameter set for one of a continuous period of one of the at least one of the next mixing channels, and Wherein the level parameter calculator (900) is operative to calculate a level parameter for each of a continuous period of one of the at least one downstream mixing channel. The apparatus of claim 1, wherein the output interface is operative to generate an adjustable data stream, the adjustable data stream including parameters of the first subgroup of the parameter set in a lower adjustment layer, the A subgroup parameter allows reconstruction of an output channel of the first subgroup, the adjustable data stream including parameters of a second subgroup of the parameter set in a higher adjustment layer, the second subgroup and the first The subgroups together allow reconstruction of the output channels of the second subgroup, and wherein the output interface is further manipulated to cause the level parameter to enter the lower adjustment layer. A device for reconstructing a multi-channel representation using a parameter representation to generate one of a raw multi-channel signal having at least three original channels, the parameter representation having a parameter set, the parameter set being A multi-channel reconstruction is allowed when at least the next mixing channel is used together. The parameter representation includes a level parameter that is a level difference between a primary downmix and a parametric downmix, wherein The parameter indicates that the downmix is based on the parameter, and the device includes: a level corrector (902) for using the level parameter to apply level correction of a single basic channel or a plurality of basic channels, using The level parameter weights the single base channel or the plurality of base channels so that an upmixing can be performed by using parameters in the parameter set to obtain a corrected multi-channel reconstruction. A method of generating a level parameter in a parameter representation of a multi-channel signal, the multi-channel signal having a plurality of original channels, the parameter representation comprising a parameter group, the parameter group being mixed with at least the next line A multi-channel reconstruction is allowed when the channels are used together, the method comprising: calculating (900) a level parameter (r _M ), the level parameter being a level between a main down mix and a parameter down mix a difference, wherein the parameter indicates that the downmix is based on the parameter; and generating an output data, the output data including the level parameter and the parameter set or the level parameter and the at least one lower mix channel. A method of reconstructing a multi-channel representation using a parameter representation to produce one of a raw multi-channel signal having at least three original channels, the parameter representation having a parameter set, the parameter set being A multi-channel reconstruction is allowed when at least the next mixing channel is used together. The parameter representation includes a level parameter that is a level difference between a primary downmix and a parametric downmix, wherein The parameter representation is based on the parameter downmixing, the method comprising: using the level parameter to implement (902) level correction of a single base channel or a plurality of base channels, using the level parameter to weight the single The basic channel or a number of basic channels so that an upmixing can be performed by using the parameters in the parameter set to obtain a corrected multi-channel reconstruction. A recording medium having a computer program with machine readable instructions, when executed on a computer, implements the method as described in claim 6 or 7.