DE602004008613T2 - TREUE OPTIMIZED CODING WITH VARIABLE FRAME LENGTH - Google Patents

Abstract

A method of encoding multi-channel audio signals comprises generating of a first output signal (x' mono ), being encoding (38) parameters representing a main signal (x mono ). The main signal (x mono ) is a first linear combination (34) of signals (16A,16B) of at least a first and a second channel. The method further comprises generating (30) of a second output signal (p side ), being encoding parameters representing a side signal (x side ). The side signal (x side ) is a second linear combination (36) of signals (16A,16B) of at least the first and the second channel within an encoding frame. The method is characterised in that the generating of the second output signal further comprises scaling of the side signal (x side ) to an energy contour of the main signal (x mono ). A method of decoding is also presented as well as an encoder, a decoder and audio system, all according to the same basic idea.

Description

TECHNICAL AREA

The The present invention relates generally to the coding of audio signals and especially the coding of multi-channel audio signals.

BACKGROUND

It there is a big one Demand on the market for the transferring and storing audio signals at low bit rate while high Audio Quality is maintained. Especially in cases in which transmission resources or memory is limited, the operation is at low bit rate an essential cost factor. This is typical of the case z. B. in power and messaging applications in mobile communication systems, such as GSM, UMTS or CDMA.

Present are no standardized encoder-decoder available, providing high stereophonic Audio Quality at bit rates, which is economical interesting for which are use in mobile communication systems. Which is possible with available Encoders / decoders, is monophonic transmission of audio signals. To some extent also stereophonic transmission available. However, bit rate limits usually require the stereo presentation drastically limit.

Of the simplest way of stereophonic or multi-channel coding of audio signals is to the signals from different channels separated as individual and independent Encode signals. Another fundamental way, used in Stereo FM radio transmission, which also has compatibility with legal mono-radio receivers saves, a sum and a difference signal of the two channels involved is to be transmitted.

Encoder / decoder of the prior art, such as MPEG-1/2 Layer III and MPEG-2/4 AAC use so-called common stereo coding. According to this technique the signals of the different channels are processed together instead of separate and individual. The two most used techniques the common stereo coding are known as "mid / side" (M / S) stereo coding and intensity stereo coding, which usually on sub-bands the stereo or multi-channel signals to be coded are used.

M / S stereo coding is similar to the procedure described for stereo FM radio in the sense that it encodes and transmits the sum and difference signals of the channel sub-bands, thereby exploiting redundancy between the sub-bands. The structure and operation of a coder based on M / S stereo coding is described e.g. In U.S. Patent 5,285,489 by JD Johnston.

On the other hand, intensity stereo coding is capable of using stereo irrelevance. It transmits the common intensity of the channels (the various sub-bands) along with some location information, indicating how the intensity is distributed among the channels. Intensity stereo coding provides only information about the spectral intensity of the channels. Phase information is not transported. For this reason and because the temporal inter-channel information (more precisely, the time difference between channels) is of greater psychoacoustic relevance, especially at lower frequencies, intensity stereo coding can only be used at high frequencies above about 2 kHz. An intensity stereo coding method is described, for. B. in the European Patent 04974413 by Veldhuis et al.

One recently developed stereo coding method is described for. In a conference publication entitled "Binaural cue coding applied stereo and multi-channel audio compression ", 112th AES convention, May 2002, Munich, Germany by C. Faller et al. This method is a procedure for parametric multi-channel audio coding. The basic principle is that on the coding side the signals of N channels c1, c2, ... cn be combined into a mono signal m. The mono signal is audio encoded, using any conventional one monophonic audio encoder / decoder. In parallel, parameters are derived from the channel signals, which describe the multi-channel image. The parameters are coded and transmitted to the decoder, along with the audio bit stream. The decoder first decodes the mono signal m 'and regenerates the channel signals c1', c2 ', ..., cN' based on the parametric description of the multi-channel image.

The principle of 'Binaural Cue Coding' (BCC) is that it is the encoded mono signal and so on transmits said BCC parameter. The BCC parameters include coded level differences and time differences between the channels for sub-bands of the original multi-channel input signal. The decoder re-generates the various channel signals by applying level and phase adjustments per sub-band based on the BCC parameters. The advantage over z. M / S or intensity stereo is that the stereo information comprising inter-channel temporal information is transmitted at much lower bit rates. However, this technique requires sophisticated time-frequency transforms for each of the channels in terms of computation, both in the encoder and in the decoder.

Furthermore BCC does not handle the fact that a lot of the stereo information, especially at low frequencies is diffuse, that means it comes not from any direction. Diffuse tone fields exist in both channels a stereo recording, but these are not to a large extent in phase in relation to each other. If an algorithm, such a like BCC, subject records with a large amount of diffuse clay fields The reproduced stereo image is confused, jumping from left to right as the BCC algorithm the signal only in certain frequency bands to the right or left can classify.

One potential Means to encode the stereo signal and good playback of To ensure diffuse sound fields, a coding scheme is very similar to in FM stereo broadcasting used, namely the mono signals (left + right) and the difference signals (left-right) to code separately.

A technique described in U.S. Patent 5,434,948 by CE Holt et al., uses a similar technique as in BCC for coding the mono signal and the page information. In this case, the page information consists of predictor filters and optionally a residual signal. The predictor filters, estimated by a least squares algorithm when applied to the mono signal, allow the prediction of multi-channel audio signals. With this technique, one is able to achieve encoding of very low bit rates from multi-channel audio sources, but at the cost of a quality drop, as discussed below.

Ultimately, for the Completeness, is a technique to mention which for 3-D audio is used. This technique synthesizes the right one and left channel signals by filtering the audio source signals so-called head-related filters. However, this technique requires that the different sound source signals are separate, and can therefore not generally for Stereo or multi-channel coding.

ZUSAMNENFASSUNG

One Problem with existing coding schemes based on coding of frames of signals, especially a main signal and one or more side signals, is that the subdivision of audio information in frame unattractive Can produce sensation artifacts. Dividing the information in a relatively long duration usually reduces the average demanded bit rate. This can be advantageous for. For music, containing a large one Amount of diffused clay. However, for music, rich in transitions or Language, fast temporal variations will be smeared over the Duration of the frame, causing ghostly sounds or also pre-echo problems. Encoding short frames will instead be a more accurate presentation of the sound, minimizing the energy but requiring higher transmission bit rates and larger calculation capacities. The Encoding efficiency as such can vary with very short frame lengths as well reduce. The introduction more frame boundaries may also introduce discontinuities in encoding parameters can appear as perceptual artifacts.

One Another problem with schemes based on coding a Main signal and one or more side signals is that these often relatively large Calculation capacity require. In particular, when short frames are used is managing discontinuities in the parameters from one frame to another a complex one Task. When long frames are used, estimation errors of tone transitions can be very high size Cause side signals, which in turn reduces the need for transmission rate increase.

One The aim of the present invention is therefore to provide a coding method and a Device to improve the perceptual quality of multi-channel audio signals, especially to avoid artifacts, such as pre-echo, ghostly Sounds or Frame discontinuity artefacts. Another object of the present invention is an encoding method and to provide a device which has lower processing power require and more constant transmission bit rate requirements to have.

The The above objects are achieved by methods and devices accordingly the appended claims. Generally words Polyphonic signals are used to produce a main signal, typical a mono signal, and a side signal to create. The main signal becomes according to coding principles of the prior art coded. A number of coding schemes for the Page signal is provided. Each coding scheme is marked by a group of sub-frames of different lengths. The whole length the sub-frame corresponds to the length of the coding frame of the Coding scheme. The groups of subframes include at least one subframe. That for the encoding scheme to be used for the page signal is selected, at least partially dependent from the present Signal content of the polyphonic signals.

In an embodiment the selection takes place before coding, based on analysis the signal properties. In another embodiment, the page signal encoded by each of the coding schemes and then being based on quality measurements coding the best coding scheme selected.

In a preferred embodiment a residual signal is generated as a difference between the Page signal and the scaled with a balance / compensation factor Main signal. The balance / compensation factor is selected to to minimize the page residual signal. The optimized side residual signal and the adjustment / compensation factor is encoded and called the page signal representing parameters provided. On the decoder side, the balance / equalization factor, the side residual signal and the main signal used to get that Restore page signal.

In a further preferred embodiment Encoding the side signal includes energy contour scaling to avoid pre-echo effects. You can also use different coding schemes different coding procedures in the separate sub-frames.

Of the Main advantage of the present invention is that the conservation the perception / sensation of the audio signals is improved. Also allowed the present invention also involves the transmission of multi-channel signals at very low bit rates.

BRIEF DESCRIPTION OF THE DRAWINGS

The Invention, together with its other objectives and advantages, can are best understood by reference to the following description, taken together with the accompanying drawings, in which:

1 shows a block diagram of a system for transmitting polyphonic signals;

2a a block diagram of an encoder in a transmitter;

2 B a block diagram of a decoder in a receiver;

3a Figure 11 is a diagram illustrating coding frames of various lengths;

3b and 3c Block diagrams of embodiments of page signal encoding units according to the present invention are;

4 Fig. 12 is a block diagram of one embodiment of an encoder using adjustment / compensation factor coding of the side signal;

5 Figure 12 is a block diagram of one embodiment of a multi-channel system encoder;

6 a block diagram of an embodiment of an encoder, suitable for decoding the signals from the device of 5 , is;

7a and 7b Charts, illustrating pre-echo artifacts, are;

8th a block diagram of one embodiment of a page signal encoding unit according to the present invention employing different coding principles in different subframes;

9 illustrates the use of different coding principles in different frequency sub-bands;

10 Fig. 10 is a flow chart of the basic steps of one embodiment of a coding method according to the present invention; and

11 Fig. 10 is a flow chart of the basic steps of one embodiment of a decoding method according to the present invention.

DETAILED DESCRIPTION

1 illustrates a typical system 1 in which the present invention can be used to advantage. A transmitter 10 contains an antenna 12 Including associated hardware and software to be able to transmit radio signals 5 to a receiver 20 to send. The transmitter 10 contains among other parts a multi-channel encoder 14 representing the signals of a number of input channels 16 in output signals, suitable for radio transmission, transformed. Examples of suitable multi-channel encoders 14 will be described in detail below. The signals of the input channels 16 can z. B. from an audio signal memory 18 , such as a data file of the digital representation of audio recordings, of magnetic tape or vinyl-disk recordings of audio, etc. The signals of the input channels 16 can also be "live", for example from a group of microphones 19 to be delivered. The audio signals are digitized, if not already in digital form, before those in the multi-channel coder 14 enter.

On the side of the recipient 20 handles an antenna 22 with associated hardware and software, the actual reception of radio signals 5 representing polyphonic audio signals. Typical functions, such as error correction, are executed here. A decoder 24 decodes the received radio signals 5 and transforms the audio data transported thereby into signals of a number of output channels 26 , The output signals can be supplied e.g. B. to speakers 29 for instant presentation or can be in an audio signal memory 28 be stored of any kind.

The system 1 may be, for example, a telephone conference system, a system for providing audio services or other audio applications. In some systems, such as. As the telephone conference system, the communication must be of a duplex type, while z. For example, the distribution of music from a service provider to a service customer may be essentially one-way. The transmission of signals from the transmitter 10 to the recipient 20 can also by any other means, for. As different types of electromagnetic waves, cables or fibers, as well as by combinations of these are performed.

2a illustrates an embodiment of an encoder according to the present embodiment. In this embodiment, the polyphonic signal is a stereo signal comprising two channels a and b, received at the inputs, respectively 16A and 16B , The signals of channels a and b are sent to a preprocessing unit 32 delivered, in which various signal conditioning procedures can be performed. The (perhaps modified) signals from the output of the preprocessing unit 32 be in the adder unit 34 summed. This adder unit 34 also divides the sum by a factor 2. The signal x _mono generated in this way is a main signal of the stereo signal, since it comprises substantially all data from both channels. Therefore, in this embodiment, the main signal represents a pure "mono" signal, and the main signal x _mono is sent to a main signal encoding unit 38 which encodes the main signal according to any suitable coding principles. Such principles are available within the state of the art and therefore will not be discussed further here. The main signal coding unit 38 provides an output signal p _mono consisting of coding parameters representing a main signal.

In a subtraction unit 36 For example, a difference (divided by a factor of 2) of the channel signals is provided as a side signal x _side . In this embodiment, the side signal represents the difference between the two channels of the stereo signal. The page signal x _side is sent to a page signal encoding unit 30 delivered. Preferred embodiments of the side signal coding unit 30 will be discussed below. In accordance with a page signal encoding procedure, which will be described in more detail below, the page signal x _{side is converted} to the encode parameters p _side representing a page signal x _side . In certain embodiments, this coding also takes place using the information of the main signal x _mono . The arrow 42 shows such a device in which the original unencoded signal x _mono is used. In other different embodiments, the main signal information contained in the page signal encoding unit 30 is derived from the coding parameters p _mono representing the main signal, as deduced by the broken line 44 displayed.

The coding parameters p _mono representing the main signal x _mono are a first output signal and the coding parameters p _side represent the side signal x _side in a second output signal. In a typical case, these two output signals p _mono , p _side , together representing the full stereo sound, become a transmission signal 52 in a multiplexer unit 40 multiplexed. However, in other embodiments, the transmission of the first and second output signals p _mono , p _{side may} take place separately.

In 2 B is an embodiment of a decoder 24 illustrated as a block scheme according to the present invention. The received signal 54 comprising coding parameters representing the main signal and side signal information, is sent to a demultiplexer unit 56 which separates a first and a second input signal, respectively. The first input signal, corresponding to the coding parameters p _{mono of} a main signal, is sent to a main signal decoding unit 64 delivered. In a conventional manner, the coding parameters p _mono representing the main signal are used to _mono generate a decoded main signal x "which is as similar as possible to the main signal x _mono ( 2a ) of the encoder 14 ( 2a ).

Similarly, the second input signal corresponding to a page signal is sent to a page signal decoder unit 60 delivered. Here, the encode parameters p _side representing the page signal are used to restore the decoded side signal x " _side . In some embodiments, the decoding procedure uses information about the main signal x " _mono , as indicated by the arrow 65 displayed.

The decoded main and side signals x " _mono and x" _side are sent to an adder unit 70 which provides an output signal which is a representation of the original signal of channel a, similarly provides one by a subtraction unit 68 provided difference an output signal, which is a representation of the original signal of the channel b. These channel signals can be post-processed in a post-processing unit 74 according to signal processing procedures of the prior art. Finally, the channel signals a and b at the outputs 26A and 26B provided by the decoder.

As mentioned in the summary, coding is typically performed at one time in a frame. A frame contains audio samples within a predefined period of time. In the lower part of the 3a a frame SF2 of the time duration L is illustrated. The audio samples within the non-dashed area are to be coded together. The preceding and following samples are encoded in different frames. The division of samples into frames will definitely create some discontinuities at the frame boundaries. Moving sounds will result in shifting of coding parameters which basically change at each frame boundary. This will give rise to perceptible errors. One way to compensate for this is to base the coding not only on the samples to be coded, but also on samples in the immediate vicinity of the frame, as indicated by the dashed areas. In this way, there will be a smoother transition between the different frames. As an alternative or supplement, interpolation techniques are sometimes used to reduce perceptual artifacts caused by framing boundaries. However, all of these procedures require large additional computational resources, and for certain special decoding techniques, it may also be difficult to provide any resources.

in view of it is advantageous as possible to use long frames, since then the number of frame boundaries will be small. Also, the coding efficiency is typically high and high the required transmission bit rate will typically be minimized. However, long frames result in problems with pre-echo artifacts and ghostly tones.

If, instead, shorter frames are used, such as SF1 or even SF0, having the time periods of L / 2 and L / 4, respectively, anyone familiar with the art will recognize that the coding efficiency can be reduced, the transmission bit rate is higher and the problems with frame boundary artifacts will grow. However, shorter frames suffer less from other perceptual artifacts, such as ghostly sounds and pre-echo. To minimize the coding error as much as possible, you should choose a frame length as short as possible.

Corresponding In the present invention, the audio perception is improved by using a Frame length for the Encoding the page signal, which depends on the currently available one Signal content. Because the influence of different frame lengths on the audio perception will be different, depending on the nature of the coded Sounds, can An improvement can be achieved by the nature of the signal itself is allowed to influence the frame length used. Coding the Main signal is not the subject of the present invention and is therefore not described in detail. However, those for coding the main signal used frame lengths equal to or not equal to those for encoding the page signal used frame lengths.

Because of small temporal changes It can in some cases useful be, the side signal using relatively long frame to encode. This can be the case with records with a large amount of diffuse clay field, such as concert records. In other cases, such as stereo-voice conversation, are probably short frames to prefer. The decision, which frame length to prefer is can be met in two basic ways.

An embodiment of a page signal encoding unit 30 according to the present invention is in 3b illustrated in which an infinite loop decision is used. An elementary coding frame of length L is used here. A number of coding schemes 81 , characterized by a separate group 80 from sub-frame 90 , are generated. Every group 80 from sub-frame 90 contains one or more sub-frames 90 of equal or different length. The entire length of the group 80 from sub-frame 90 is however always equal to the elementary coding frame length L. With references to 3b For example, the uppermost coding scheme is characterized by a group of subframes containing only one subframe of length L. The next group of sub-frames contains two frames of length L / 2. The third group contains two frames of length L / 4, followed by an L / 2 frame.

The signal x _side , supplied to the side signal coding unit 30 is through all coding schemes 81 coded. In the top encoding scheme, the entire elementary frame is encoded in one piece. However, in the other schemes, the signal x _side in each subframe is coded separately from each other. The result from each coding scheme is sent to a selector 85 delivered. A fidelity measuring means 83 determines a fidelity measure for each of the encoded signals. The fidelity measure is an objective quality value, preferably a signal-to-noise measure or a weighted signal-to-noise ratio. The fidelity measurements associated with each coding scheme are compared and the result controls a switching means 87 to the coded parameters representing the page signal from the coding scheme which has given the best fidelity measure, as the output signal p _side from the page signal encoder unit 30 select.

Prefers be all possible Combinations of frame lengths tested and the Group of sub-frames that gives the best objective quality, z. As signal-to-noise ratio is selected.

wobei l_sf die Längen der Unter-Rahmen, l_f die Länge des Kodier-Rahmens und n ein Integer ist. In der vorliegenden Ausführungsform wird n zwischen 0 und 3 gewählt, es wird möglich sein irgendeine Rahmen-Länge zu verwenden, so lange wie die gesamte Länge der Gruppe konstant gehalten wird.In the present embodiment, the lengths of the sub-frames used are selected according to:

where l _{sf is} the lengths of the subframes, l _{f is} the length of the encoding frame, and n is an integer. In the present embodiment, n is selected between 0 and 3, it will be possible to use any frame length as long as the entire length of the group is kept constant.

In 3c is another embodiment of a page signal encoder unit 30 illustrated in accordance with the present invention. Here, the decision of frame length is an infinite loop decision based on the statistics of the signal. In other words, the spectral properties of the page signal are used as a basis for deciding which coding scheme to use. As before, various coding schemes characterized by different groups of subframes are available. In this embodiment, the selector 85 however, placed before actual coding. The input side signal x _side enters the selector 85 and a signal analyzing unit 84 one. The result of the analysis is the input of a switch 86 in which only one of the coding schemes 81 is used. The output from this coding scheme will also be the output signal p _side of the _page th signal coding unit 30 be.

The advantage with an infinite loop decision is that only one coding is actually to be executed. The disadvantage, however, is that the analysis of the signal characteristics can actually be very complicated and it can be difficult to predict possible behavior to be able to be in the switch 86 to make a suitable choice. A lot of statistical analysis of sound needs to be done and in the signal analysis unit 84 be included. Any little change in the coding schemes can turn statistical behavior upside down.

By using the infinite loop selection ( 3b ) coding schemes may be exchanged without any other change in the remainder of the unit. On the other hand, if many coding schemes are to be investigated, the computational power requirements will be high.

Of the The benefit of such variable length coding for the page signal is that between a fine temporal resolution and coarse frequency resolution one side and gross temporal resolution and fine frequency resolution chosen the other side can be. The above embodiments will be the stereo picture in the best possible Received manner.

It There are also some requirements for the actual coding used in the various coding schemes. Especially, if the infinite loop selection is used, the Computing power resources be great to perform a number of more or less simultaneous coding. more the more complex the coding process, the more computational power is required. Furthermore a low bit rate is also preferable in the transmission.

This in US 5,434,948 The presented method uses a filtered version of the mono (main) signal to equal the page or difference signal. The filter parameters are optimized and may vary over time. The filter parameters are then transmitted representing a coding of the page signal. In one embodiment, a page residual signal is also transmitted. In many cases, such an approach would be possible to be used as the page signal encoding method within the scope of this invention. However, this procedure has some disadvantages. The quantization of the filter coefficients and any side residual signal often require relatively high bit rates for transmission because the order of the filter must be high to provide an accurate estimate of the side signal. The estimation of the filter itself can be problematic, especially in cases of transient-rich music. Estimation errors will result in a modified side signal, which is sometimes greater in magnitude than the unmodified signal. This will lead to higher bit rate requirements. In addition, when calculating a new set of filter coefficients after every N samples, the filter coefficients must be interpolated to give a smooth transition from one set of filter coefficients to another, as discussed above. Interpolation of filter coefficients is a complex task and errors in the interpolation will manifest themselves in large side-error signals, leading to higher bit-rates needed for the difference-error-signal-coder.

One Means to the need for To avoid interpolation, the filter coefficients are on one Sample-by-sample basis to update and to backward / adaptable Trust analysis. For This work will require that the bit rate of the residual encoder be pretty much is high. This is therefore not a good alternative for stereo coding at low bit rate.

It there are cases z. B. quite common with music in which the mono and differential signals are almost are uncorrelated. The filter estimation then becomes very painstaking the additional Risk things for the difference error signal encoder just to make it worse.

The solution according to US 5,434,948 can work quite well in cases where the filter coefficients change very slowly over time, e.g. B. in telephone conference systems. In the case of music signals, this approach does not work very well, since the filters have to change very fast to follow the stereo image. This means that subframe lengths of very different sizes are to be used, which means that the number of combinations to be tested increases rapidly. This in turn means that the requirements for computing all possible coding schemes become impractically high.

Therefore, in a preferred embodiment, the coding of the side signal is based on the idea of reducing the redundancy between the mono and the side signal by using a simple balance / equalization factor instead of a complex bit rate consuming predictor filter. The Resi duum of this process is then coded. The size of such residual is relatively small and does not require high bit rate for transmission. In fact, this idea is very well suited to combine with the variable frame group approach described earlier, since the computational complexity is low.

The Using an adjustment / compensation factor combined with the Variable frame length approach eliminates the need for complex interpolation and the related problems, which may cause interpolation. Furthermore results in the use of a simple adjustment / compensation factor instead of a complex filter, there are fewer problems with the estimation because possible estimation errors for the Balance / compensation factor have less influence. The preferred solution will be able both, shifted signals and diffuse sound fields, with good quality and with limited bit rate requirements and computational resources too reproduce.

4 illustrates a preferred embodiment of a stereo encoder according to the present invention. This embodiment is very similar to that in FIG 2a shown, however, with the details of the side signal coding unit 30 shown. The encoder 14 This embodiment has no preprocessing unit and the input signals are sent directly to the addition and subtraction units 34 . 36 delivered. The mono signal x _mono becomes a multiplier with a certain equalize / equalize factor g _sm 33 multiplied. In a subtraction unit 35 For example, the multiplied mono signal is subtracted from the side signal x _side , which is mainly the difference between the two channels to produce a residual signal. The balance factor g _sm is determined based on the content of the mono and side signals by the optimizer 37 to minimize the residual signal according to a quality criterion. The quality criterion is preferably a least squares criterion. The page residual signal is in a page residual encoder 39 correspond to any coding procedure coded. Preferred is the page residual encoder 39 a low-bit-rate-transform encoder or a codebook excited linear prediction (CELP) encoder. The coding parameters p _side representing the page signal then contain the coding parameters p _{side residual} , representing the side residual signal and the optimized tuning / equalizing factor 49 ,

In the embodiment of the 4 is the mono signal 42 , for synthesizing the page signals, uses the x _mono signal for the mono encoder 38 , As mentioned above (in conjunction with 2a ) can be the local synthesis signal of the mono-coder 38 also be used. In the latter case, the overall encoder delay may be increased and the computational complexity for the page signal may increase. On the other hand, the quality can be better, since it is then possible to repair coding errors made in the mono-coder.

In a more mathematical manner, the basic coding scheme can be described as follows. Designate the two channels as a and b, which can be the left and right channels of a stereo pair. The channel signals are combined by addition into a mono signal and by subtraction into a side signal. In equation form the operations are described as: x mono (n) = 0.5 (a (n) + b (n)) x side (n) = 0.5 (a (n) - b (n))

It is useful to scale the x _mono and x _side signals down by a factor of 2. It is assumed here that other ways of generating x _mono and x _side exist. For example, you can use: x mono (n) = γa (n) + (1-γ) b (n) x side (n) = γa (n) - (1-γ) b (n) 0 ≤ γ ≤ 1.0.

For blocks of input signals, a modified residual side signal is calculated according to: x side residual (n) = x side (n) - f (x mono' x side ) x mono (N) where f (x _mono , x _side ) is an equalize / equalize factor function which is on the block of N samples, the means a sub-frame based on the page and mono signal and which strives to remove as much of the page signal as possible. In other words, the balance / equalization factor is used to minimize the side residual signal. In the particular case where it is minimized in terms of least error squares, this is equivalent to minimizing the energy of the side residual signal x _{side residual} .

wobei x_side das Seiten-Signal und x_mono das Mono-Signal ist. Bemerke, dass die Funktion basiert auf einem Block, beginnend bei „frame start" (Rahmen-Beginn) und endend bei „frame end" (Rahmen-Ende).In the special case mentioned above, f (x _mono , x _side ) is described as:

where x _{side is} the side signal and x _{mono is} the mono signal. Note that the function is based on a block, starting at "frame start" and ending at "frame end".

It is possible to add weighting in the frequency domain when calculating the adjustment / compensation factor. This is done by convolution of the x _side and x _mono signals with the impulse response of a weighting filter. It is then possible to shift the estimation errors to a frequency range where they are less easy to hear. This is called perceptual weighting.

A quantized version of the trim / offset factor value given by the function f (x _mono , x _side ) is transmitted to the decoder. It is preferable to already consider the quantization when the modified signal is generated. The following expression is then obtained: x side residual (n) = x side (n) - g Q x mono (N)

Q _g (..) is a quantization function which is applied to the balance / equalization factor given by the function f (x _mono , x _side ). The balance / equalization factor is transmitted in the transmission channel. In normal left-right signals, the adjustment / compensation factor is limited to the interval [-1.0 1.0]. On the other hand, if the signals are out of phase with each other, the balance / equalization factor may go beyond these limits.

wobeiAs an optional means to stabilize the stereo image, one can limit the balance / equalization factor if the normalized cross-correlation between the mono and the side signal is poor as given by the following equation:

in which

These situations occur quite often at z. Classical music or studio music with a large proportion of diffused sounds, where the a and b channels in some cases almost cancel each other out in cases where a mono signal is generated. The effect on the balance / equalization factor is that this can jump quickly, creating a blurred stereo image. The above correction weakens this Problem.

The filter-based approach in US 5,434,948 has similar problems, but in that case the solution is not so easy.

If E _{s is} the coding function (eg, a transform coder) of the side residual signal and E _{m is} the coding function of the mono signal, then the decoded a "and b" signals in the decoder -End be described as (it is assumed here that γ = 0.5): a '' (n) = (1 + g Q ) X '' mono (n) + x '' side (N) b '' (n) = (1 - g Q ) X '' mono (n) - x '' side (N) x '' side = E s -1 (e s (x side residual ) x '' mono = E m -1 (e m (x mono ))

One important benefit from calculating the balance / compensation factor for each The framework is that the use of interpolation is avoided. Instead, normally, as described above, the frame processing is overlapping Frame executed.

The Using Coding Principle Matching / Compensation Factors works especially good in the case of music signals, where typically fast change needed to follow the stereo picture.

Recently, multi-channel coding has become popular. An example is 5.1ch surround sound in DVD movies. The channels are then arranged like: front left, front center, front right, back left, back right and sub woofer. In 5 For example, an embodiment of an encoder is illustrated that encodes the three front channels in such an arrangement utilizing inter-channel redundancies according to the present invention.

The three channels L, C, R are connected to the three inputs 16A -C and the mono signal x _mono is generated as the sum of all three channels. A center signal encoder unit 130 is added, which receives the center signal x _center . The mono signal 42 In this embodiment, the coded and decoded mono signal x "is _mono and is multiplied by a certain equalize / equalize factor g _Q in a multiplier 133 , In a subtracting unit 135 The multiplied mono signal is subtracted from the center signal X _center to produce a center residual signal. The balance / compensation factor g _Q is determined based on the contents of the mono and center signals by an optimizer 137 to minimize the center residual signal according to a quality criterion. The center signal is in a center-to-field encoder 139 encoded according to any coding procedure. Preferred is the center-to-center encoder 139 a low bit rate transform encoder or a CELP encoder. The coding parameters p _center , representing the center signal, then comprise the coding parameters p _{center residual} , representing the _{center residual} signal and the optimized tuning / equalizing factor 149 , The center residual signal and the scaled mono signal are stored in an adder unit 235 adds generating a modified center signal 142 , which is compensated for coding errors.

The side signal x _side , that is the difference between left L and right R channels, is sent to the side signal coder unit 30 as supplied in previously discussed embodiments. However, here's the optimizer 37 also from the modified center signal 142 supplied by the center signal encoder unit 130 , from. The side residual signal is therefore considered to be an optimal linear combination of the mono signal 42 , the modified center signal 142 and the side signal in the subtraction unit 35 generated.

The The concept of variable frame lengths described above can be used on any of the and center signals or both.

6 illustrates a decoding unit useful for receiving coded audio signals from a coding unit of the present invention 5 , The received signal 54 is divided into encode parameter p _mono representing the main signal, encode parameter p _center representing the center signal and encode parameter p _side representing the page signal. In the decoder 64 For example, the coding parameters p _mono representing the main signal are used to _mono generate a main signal x ". In the decoder 160 become the Ko the parameter p _center representing the center signal used to generate a center signal x " _center based on the main signal x" _mono . In the decoder 60 the coding parameters p _side representing the page signal are decoded generating a page signal x " _side , based on the main signal x" _mono and the center signal x " _center . The procedure can be expressed mathematically as follows:
The input signals x _left , x _right and x _center are combined into a mono channel corresponding to: x mono (n) = αcx left (n) + βx right (n) + χx center (N) α, β and χ are set to 1.0 in the remaining section for simplicity, but they can be set to any values. The α, β and χ values can either be constant or dependent on the signal content to highlight one or the other channel for optimum quality.

wobeiThe normalized cross-correlation between the mono and center signals is calculated as:

in which

x _{center is} the center signal and x _{mono is} the mono signal. The mono signal comes from the mono-target signal, but it is equally possible to use the local synthesis of the mono-coder.

Q_g(..) ist eine Quantelung-Funktion, welche auf den Abgleich/Ausgleich-Faktor angewandt ist. Der Abgleich/Ausgleich-Faktor wird über den Übertragungs-Kanal übertragen.The center residual signal to be encoded is: x center residual (n) = x center (n) - g Q x mono (N)

Q _g (..) is a quantization function applied to the balance / equalization factor. The adjustment / compensation factor is transmitted via the transmission channel.

If E _{c is} the coding function (eg, a transform coder) of the center residual signal and E _{m is} the coding function of the mono signal, then the decoded x _center signal at the end of the decoder may be described are as: x '' center (n) = g Q x '' mono (n) + x '' center residual (N) x '' center residual = E c -1 (e c (x center residual )) x '' mono = E m -1 (e m (x mono )).

η kann zum Beispiel gleich 2 sein für eine Fehler-Minimierung der kleinsten Fehler-Quadrate. Die g_sm und g_sc-Parameter können gemeinsam oder getrennt gequantelt sein.The page residual signal to be encoded is: x side residual (n) = (x left (n) - x right (n)) - g Qsm x '' mono (n) - g qsc x '' center (N) where g _Qsm and g _{Qsc are} quantized values of the parameters g _sm and g _sc which minimize this expression:

For example, η may equal 2 for an error minimization of the smallest error squares. The g _sm and g _sc parameters can be shared or separated.

If E _{s is} the encoding function of the side residual signal, then the decoded x '' _left and x '' _right channel signals are given as: x '' left (n) = x '' mono (n) - x '' center (n) + x '' side (N) x '' right (n) = x '' mono (n) - x '' center (n) + x '' side (N) x '' side (n) = x '' side residual + g Qsm x '' mono (n) + g qsc x '' center (N) x '' side residual = E s -1 (e s (x side residual )).

One of the most disturbing perceptual artifacts is the pre-echo effect. In the 7a -b diagrams illustrate such an artifact. Suppose a signal component having the temporal evolution as through the curve 100 shown. At the beginning, starting with t0, the signal component is not present in the audio sample. At a time t between t1 and t2, the signal component suddenly appears. If the signal component is coded using a frame length of t2-t1, the occurrence of the signal component will be "smeared" over the entire frame, as in curve 101 displayed. If a decoding of the curve 101 takes place, the signal component appears at a time Δt before the intended appearance of the signal component, and a "pre-echo" is detected.

The Pre-echo artifacts become more apparent when using long encoding frames become. By using shorter ones Frame, the artifact is somewhat suppressed. Another way to go up to treat pre-echo problems described is the fact too that the mono signal is available on both the encoder and and the decoder end. This makes it possible to adjust the page signal scale, according to the energy contour of the mono signal. At the decoder end inverse scaling is done and so can some the pre-echo problems be relieved.

wobei w(n) eine Fenster-Funktion ist. Die einfachste Fenster-Funktion ist eine Rechtecks-Funktion, es können aber andere Fenster-Typen, solche wie ein Hamming-Fenster, bevorzugter sein.An energy contour of the mono signal is calculated over the entire frame as:

where w (n) is a window function. The simplest window function is a rectangle function, but other types of windows, such as a Hamming window, may be more preferable.

The page residual signal is then scaled as:

wobei f(..) eine monotone kontinuierliche Funktion ist. In dem Dekodierer wird die Energie-Kontur für das dekodierte Mono-Signal berechnet und auf das dekodierte Signal angewandt als: x''side(n) = x''side(n)f(Ec(n)), frame start ≤ n ≤ frame end. In a more general form, the above equation can be written as:

where f (..) is a monotonic continuous function. In the decoder, the energy contour for the decoded mono signal is calculated and applied to the decoded signal as: x '' side (n) = x '' side (N) f (E c (n)), frame start ≤ n ≤ frame end.

Since this energy contour scaling is in some ways alternative to the use of the shorter frame lengths, this concept is particularly well suited to be combined with the concept of variable frame lengths described above. By having some coding schemes that apply energy contour scaling, some that do not, and some that apply energy contour scaling only during certain sub-frames available, a more flexible set of coding may be available. Schemas are provided. In 8th is an embodiment of a signal encoder unit 30 illustrated in accordance with the present invention. Here are the different coding schemes 81 dashed sub-frame 91 representing coding applying contour scaling, and not dashed Un ter-frame 92 , representing coding procedures not applying contour scaling. In this way, combinations are available not only from subframes of different lengths, but also from subframes with different coding principles. In the present illustrative example, the application of energy contour scaling differs for different coding schemes. In a more general case, any coding principles may be analogously combined with the concept of variable lengths.

The group of coding schemes of 8th includes schemes which z. B. handle pre-echo artifacts in various ways. In some schemes, longer sub-frames are used with pre-echo minimization according to the energy contour principle. Other schemes use shorter subframes without energy contour scaling. Depending on the signal content, one of the alternatives may be more advantageous. For cases of very massive pre-echoes, coding schemes using short sub-frames with energy contour scaling may be needed.

The suggested solution may be in the entire frequency band or in one or more specific ones Sub-bands be used. The use of sub-bands can be used either on both main and side signals or separately on one of these. A preferred embodiment comprises a splitting of the side signal in some frequency bands. the reason is simple, that it is easier, the possible redundancy to remove in an isolated frequency band than in the whole Frequency band. This is especially important if music signals be encoded with rich spectral content.

A possible Application is, the frequency band below a predetermined threshold encode with the above method. the predetermined threshold may preferably be at 2 kHz or even better at 1 kHz. For the remaining Part of the frequency range of interest may be another additional one Frequency band can be encoded with the above method or it can a completely other method can be used.

It is a motivation for using the above method, preferred for low frequencies, that the diffuse tone fields generally have little energy content at high frequencies. The natural reason is that sound absorption typically increases with frequency. Also, the diffuse sound-field components at high frequencies seem to play a less important role in the human auditory system. Therefore, it is useful to use this solution at low frequencies (below 1 or 2 kHz) and rely on other, even more efficient coding schemes at higher frequencies. The fact that the scheme is applied only at low frequencies gives a great bit rate saving since the required bit rate in the proposed method is proportional to the required bandwidth. In most cases, the mono-coder can encode the entire frequency band, while the proposed coding of the page signal is suggested to be performed only in the lower part of the frequency band, as by 9 illustrated. The reference number 301 refers to a coding scheme for the page signal according to the present invention. reference number 302 refers to any other coding scheme of the page signal and reference number 303 refers to a coding scheme of the page signal.

It there is also the possibility the proposed procedure for some specific frequency bands to use.

In 10 The main steps of one embodiment of a coding method according to the present invention are illustrated as a flow chart. The procedure begins in step 200 , In step 210 a main signal derived from the polyphonic signals is coded. In step 212 Coding schemes are provided which include sub-frames of different lengths and / or order. One in the step 214 Side signal derived from the polyphonic side signals is encoded by a coding scheme selected at least in part depending on the actual signal content of the polyphone signals. The procedure ends in step 299 ,

In 11 For example, the main steps of one embodiment of a decoding method according to the present invention are illustrated as a flow chart. The procedure begins in step 200 , In step 220 a received coded main signal is decoded. In step 222 Coding schemes are provided which include sub-frames of different lengths and / or order. A received page signal is in step 224 decoded by a selected coding scheme. In step 226 The decoded main and side signals are combined into a polyphonic signal. The procedure ends in step 299 ,

The Embodiments described above are to be understood as a few illustrative examples of the present Invention. It will be understood by those familiar with the field, that different modifications, combinations and changes on the embodiments can be done without to deviate from the scope of the present invention. In particular, you can different part solutions in the various embodiments combined in other configurations where technically possible. Of the Scope of the present invention is defined, however by the appended claims.

REFERENCES

• European Patent 0497413
• US patent 5,285,498
• US patent 5,434,948
• "Binaural cue coding to stereo and multi-channel audio compression ", 112th AES convention, May 2002, Munich, Germany by C. Faller et al.

Claims (26)

A method of encoding multi-channel audio signals comprising the steps of: generating ( 210 ) of a first output signal (p _mono ), which are coding parameters representing a main signal (x _mono ), the main signal (x _mono ) being a first linear combination of signals from at least a first and a second channel (a, b; L, R); and generating ( 214 ) Of a second output signal (p _side), which are a side signal (x _side) representing encoding parameters, wherein the side signal (x _side) being a second linear combination of signals of at least a first channel and a second Channel (a, b; L, R) within a coding frame ( 80 ), characterized by the further step of: providing ( 212 ) of at least two coding schemes ( 81 ), wherein each of the at least two coding schemes is characterized by an associated group of sub-frames ( 90 ), where each group of subframes encodes the coding frame ( 80 ), the sub-frames ( 90 ) of the subframe groups have different lengths; the sum of the lengths of subframes ( 90 ) in each coding scheme ( 81 ) is equal to the length of the coding frame ( 80 ); each group of subframes ( 90 ) at least one subframe ( 90 ); wherein the step of generating ( 214 ) of the second output signal (p _side ), the step of selecting a coding scheme ( 81 ), at least in part dependent on a present signal content of the page signal (x _side ); the second output signal (p _side ) in each of the subframes ( 90 ) of the selected group of subframes ( 90 ) is coded separately. A method according to claim 1, characterized in that the step of generating ( 214 ) of the second output signal (p _side ), in turn, comprises the steps of: generating coding parameters representing the side signal (x _side ) separately within all subframes ( 90 ) of each of the at least two groups of subframes ( 90 ); Calculating a fidelity measure for each of the at least two coding schemes ( 81 ); and selecting the coded signal from the coding scheme ( 81 ) having the best fidelity measure, as the side parameter representing coding parameter (p _side ). A method according to claim 2, characterized that the fidelity mail is based on a signal-to-noise measure. A method according to any one of claims 1 to 3, characterized in that the sub-frames ( 90 ) Have lengths l _sf , corresponding to

where l _{f is} the length of the coding frame ( 80 ) and n is an integer. A method according to claim 4, characterized that n is less than a predetermined value. A method according to claim 5, characterized in that at least two coding schemes ( 81 ) comprise all permutations of subframe lengths. A method according to any one of claims 1 to 6, characterized in that the steps of generating ( 210 ) of the first output signal (p _mono ), in turn, comprise the steps of: generating the main signal (x _mono ); and coding the main signal (x _mono ) into the main signal representing coding parameters (p _mono ); the steps of creating ( 214 ) Of the second output signal in turn comprising the steps of: creating a side residual signal (x _{side residual)} as a difference between the side signal (x _side) and the main signal (x _mono) scaled by a balance / Balancing factor (g _sm ), wherein the balancing / balancing factor (g _sm ) is set as a factor minimizing the side residual signal in accordance with a quality criterion; and encoding the side residual signal and the balance / compensation factor (g _sm ) into the side signal representative coding parameter (p _side ). A method according to claim 7, characterized that the quality criterion on a measure of the smallest averaged squares is. A method according to any one of claims 1 to 8, characterized in that the step of encoding the side signal further comprises the step of: scaling the side signal (x _side ) into an energy contour of the main signal (x _mono ). A method according to claim 9, characterized in that the scaling of the side signal (x _side ) is a division by a factor which is a monotonous continuous function of the energy contour of the main signal (x _mono ). A method according to claim 10, characterized that the monotone continuous function has a square root function is. A method according to claim 10 or 11, characterized in that the energy contour, E _c , of the main signal, x _mono , is calculated via a sub-frame, corresponding to:

where L is an arbitrary factor, n is a summation index, m is the sample within the sub-frame, and w (n) is a window function. A method according to claim 12, characterized the window function is a rectangular window function. A method according to claim 12, characterized that the window function is a Hamming window function. A method according to any one of claims 1 to 14, characterized in that at least two coding schemes ( 81 ) comprise different coding principles of the side signal (x _side ). A method according to claim 15, characterized in that at least one first coding scheme of the at least two coding schemes ( 81 ) a first coding principle for the side signal (x _side ) for all sub-frames ( 90 ) and at least one second coding scheme of the at least two coding schemes ( 81 ) a second coding principle for the side signal (x _side ) for all sub-frames ( 90 ). A method according to claim 15 or 16, characterized in that at least one coding scheme of the at least two coding schemes ( 81 ) comprises the first encoding principle for the side signal (x _side ) for a subframe and the second encoding principle for the side signal (x _side ) for another subframe. A method according to claim 1, characterized in that the step of generating ( 214 ) of the second output signal (p _side ) in turn comprises these steps: analyzing the spectral characteristic of the side signal (x _side ); Select a group of subframes ( 90 ), based on the analyzed spectral characteristic; Encoding the side signal (x _side ) separately within all sub-frames ( 90 ) of the selected group of subframes ( 90 ). A method according to any one of claims 1 to 18, characterized in that the step of generating ( 214 ) of a second output signal (p _side ) in a limited frequency band. A method according to claim 19, characterized in that the step of generating ( 214 ) of a second output signal (p _side ) is only applied for frequencies below 2 kHz. A method according to claim 20, characterized in that the step of generating ( 214 ) of a second output signal (p _side ) is applied only for frequencies below 1 kHz. A method according to any one of claims 1 to 21, characterized in that the multi-channel audio signals are music signals represent. A method of decoding multi-channel audio signals, comprising the steps of: decoding ( 220 ) of coding parameters (p _mono ) representing a main signal (x _mono ) into a decoded main signal (x '' _mono ); wherein the main signal (x _mono ) is a first linear combination of signals from at least a first and a second channel (a, b; L, R); Decode ( 224 ) encoding parameters (p _side ) representing a decoded side signal (x _side ) into a decoded side signal (x '' _side ); wherein the side signal (x _side ) _comprises a second linear combination of signals from at least a first and a second channel (a, b, L, R) within a sub-frame ( 80 ); and Combine ( 226 ) of at least the main decoded signal (x " _mono ) and the decoded side signal (x" _side ) into signals from at least the first and second channels (a, b; L, R) characterized by this step : Provide ( 222 ) of at least two coding schemes ( 81 ), each of the at least two coding schemes ( 81 ) characterized by an associated group of subframes ( 90 ), where each group of subframes encodes the coding frame ( 80 ) forms; sub-frame ( 90 ) of the groups of subframes have different lengths; the sum of the lengths of subframes ( 90 ) in each coding scheme ( 81 ) is equal to the length of the coding frame ( 80 ); each group of subframes ( 90 ) at least one subframe ( 90 ); and wherein the step of decoding ( 224 ) the encoding parameter (p _side ) representing the side signal again comprises the step of decoding the coding parameters (p _side ) which separate the page signal into the subframe (p _side ). 90 ) of one of the at least two coding schemes ( 81 ). A coding device ( 14 ), comprising: input means ( 16 ; 16A -C) for multi-channel audio signals (a, b; L, R, C) comprising at least a first and a second channel (a, b, L, R); Medium ( 38 ) for generating a first output signal (p _mono ), which are coding parameters representing a main signal (x _mono ), the main signal (x _mono ) being a first linear combination of signals from at least the first and the second Channel (a, b; L, R); Medium ( 30 ) (For generating a second output signal p _side), which are a side signal (x _side) representing encoding parameters, wherein the side signal (x _side) being a second linear combination of signals of at least the first and second Channel (a, b; L, R) within a coding frame ( 80 ); and output means ( 52 ); characterized by means for providing at least two coding schemes ( 81 ), each of the at least two coding schemes ( 81 ) characterized by an associated group of subframes ( 90 ), where each group of subframes encodes the coding frame ( 80 ), sub-frames ( 90 ) of the groups of subframes have different lengths; the sum of the lengths of subframes ( 90 ) in each coding scheme ( 81 ) is equal to the length of the coding frame ( 80 ); each group of subframes ( 90 ) at least one subframe ( 90 ); where the means ( 30 ) for generating the second output signal (p _side ) again means ( 86 ; 87 ) for selecting a coding scheme which depends, at least in part, on a current signal content of the page signal (x _side ); Means for coding the side signal (x _side ) separated in each of the subframes ( 90 ) of the selected coding scheme. The decoding device ( 24 ), comprising: input means ( 54 ) for coding parameters (p _mono ) representing a main signal, and coding parameters (p _side ) representing a side signal, the main signal (x _mono ) comprising a first linear combination of a first and a second channel (p _mono ) a, b; L, R); the side signal (x _side ) is a second linear combination of a first and a second channel (a, b, L, R); Medium ( 64 ) for decoding the coding signal (p _mono ) representing the main signal into a decoded main signal (x '' _mono ); Medium ( 60 ) for decoding the page signal within a coding frame ( 80 ) encoding parameter (p _side ) into a decoded side signal (x '' _side ); Medium ( 68 . 70 ) for combining at least the main decoded signal (x " _mono ) and the decoded side signal (x" _side ) into signals from at least a first and a second channel (a, b; L, R); and output means ( 26 ; 26A -C), characterized in that the means ( 60 ) for decoding the encoding parameter (p _side ) representing the side signal, in turn comprises means for providing at least two coding schemes ( 81 ), each of the at least two coding schemes ( 81 ) characterized by an associated group of subframes ( 90 ), where each group of subframes encodes the coding frame ( 80 ), sub-frames ( 90 ) of the groups of subframes have different lengths; the sum of the lengths of subframes ( 90 ) in each coding scheme is equal to the length of the coding frame ( 80 ); each group of subframes ( 90 ) at least one subframe ( 90 ); and the means for decoding the encoding parameter (p _side ) representing the page signal is arranged for decoding the encoding parameters (p _side ) which separate the page signal into the subframe (p _side ). 90 ) of one of the at least two coding schemes ( 81 ). An audio system ( 1 ) comprising at least one of them: a coding device ( 14 ) according to claim 24, and a decoding device ( 24 ) according to claim 25.