EP1230827B1 - Method and device for processing a stereo audio signal - Google Patents

Abstract

In a device for processing a stereo audio signal having a first channel and a second channel the stereo signal is at first analyzed to obtain a measure for a quantity of bits required by a coder to code the stereo audio signal using a coding algorithm. The first channel and the second channel are then modified when the measure for the quantity of bits is larger than a predetermined value, the modification being performed in such a way that the energy of a sum signal of the first and the second modified channel is in a predetermined relation to the energy of a sum signal of the first and the second channel and that a difference signal of the first and the second modified channel is attenuated in contrast to the difference signal of the first and the second channel. Especially for audio coders requiring a constant output bit rate the side channel is attenuated in the case of stereo audio signals, the coding of which cannot meet the output bit rate of the coder, by which a stereo channel separation is abandoned for the benefit of an increased audio bandwidth or a reduction of quantizing disturbances, respectively.

Description

The present invention relates generally to coding of audio signals and in particular processing of stereo signals.

A stereo signal comprises at least two channels, i.e. H. one left channel and a right channel. In addition, you can Stereo signals still a left and right surround channel to have. There is also the possibility that a stereo signal has five different channels, d. H. a front left Channel, a front center channel and a front right channel and a left rear and a rear right channel.

For data-reducing coding of stereo signals the possibility of similarities of at least two Channels are used to reduce the amount of bits, which is needed to have a stereo signal with at least to encode two channels.

A known method for processing stereo signals in order to achieve more efficient coding is referred to as the center / side method (M / S method). The M / S method combines the first and second channels to create a center channel and a side channel. For reasons of clarity, the following will no longer speak of a first and a second channel, but of a left channel (L channel) and a right channel (R channel). It is known that the center channel is equal to the sum of the left channel L and the right channel R multiplied by a factor of 0.5, while the side channel is the difference between the left channel L and the right channel R. , multiplied by a factor of z. B. 0.5 (other factors are also possible). Expressed equally, this means: M = 0.5 · (L + R) S = 0.5 · (L - R)

If the left channel L and the right channel R are relatively similar to each other, so M / S processing brings one significant savings in the amount of bits required for coding, because the side channel is relatively less energy than R or L will have. In the borderline case, where the left channel L and the right channel R are identical, the center channel becomes the same the left channel L or the right channel R, while the side channel is 0. It can be seen that thus due to the fact that the side channel is 0, one theoretically maximum bit rate savings when coding 50% is reached because only the center channel is encoded must while no bits are spent on the side channel Need to become.

So there is a general rule that the more similar the right and left channels are the smaller, i. H. less energy, the side channel will be, and all the less Bits are required to encode the side channel.

A listener will see the similarity of the left and right channels perceive that, in the case of identical channels, a speaker or orchestra right in the middle between is perceived by the two speakers. on the other hand a listener will perceive dissimilar channels in that he has a pronounced stereo effect, d. H. that a speaker, an orchestra or individual instruments of an orchestra can be located exactly to the left and / or right can. Consider the case that the left channel a lot Has energy, and that the right channel has little energy, d. H. the case where z. B. only a single instrument is located on the far left of the recording room, and only in left channel is audible while on the right channel only Is noise, so the center channel becomes one M / S processing should be about the same as the left channel. Furthermore the side channel will also be about the same as the left Channel. In this case, both have the middle channel as well as the side channel almost the same amount of energy and both need to have a relatively large number of bits be encoded. Compared to the original case with this signal constellation, the coding required Bit size not reduced due to the M / S coding, but in the borderline case it even doubles, if of that it is assumed that the left channel L a certain amount of energy comprises, while the right channel R is 0. Here would be it was much cheaper, no M / S processing but only L / R processing. The Impact on the number of bits that are needed to encode a stereo signal is therefore sufficient in one Extreme case from a saving of 50% to another Extreme case, which requires a doubling of the coding Bits. Therefore, when using a M / S process to pay close attention to whether the piece is suitable for M / S processing or not. By doing Case in which a stereo signal (e.g. a test section of 20 ms, which is also called a frame) not for one M / S processing is appropriate for bit efficiency reasons M / S processing is dispensed with, and both the left and the right channel are coded individually. This "normal" case is also called L / R processing designated.

Usual audio coding methods, such as those used, for example in order to encode audio signals which are in accordance with one of the MPEG standards are generally decoded several steps. First, an audio signal, for example is in the form of PCM samples as they are z. B. outputs a CD player by means of a time-frequency transformation or a filter bank in a spectral representation transferred. Typically, a block with a certain number of samples, also called a "frame" is used to make a block of complex Generate spectral values that a short-term spectrum of Form frames of audio samples. Block formation is made using transformation windows reached, which are, for example, 1024 samples long. If, for example, overlapping windows for transformation are used whose degree of overlap is 50% 1024 spectral values are formed from 1024 samples. This Spectral values are then determined using a known Iteration process, after which the quantized Spectral values of entropy coding e.g. B. using a plurality of fixed Huffmann code tables subjected to finally get a bit stream on the one hand the coded quantized spectral values contains and on the other hand also page information has, which relates to the windows, to scale factors, the be calculated when quantizing and for further information which is required to decode the bit stream become.

A mid / side processing can either be done before the transformation be performed in the spectral range, d. H. using the digital time discrete samples. Alternatively, center / side processing can also be used after the transformation, d. H. with the complex spectral values be performed. The latter alternative offers furthermore the advantage that a middle / side processing not, as in the time domain, used for the entire spectrum but also for certain frequency bands, if certain spectral values of a middle / side processing undergo and others do not.

Audio coders are usually designed in such a way that they a constant bit rate, i.e. H. a certain number of bits per second. Another constraint is that the quantization noise introduced by quantization if possible, be chosen such that its Energy below the psychoacoustic masking threshold or Monitoring threshold of the audio signal is. The basic process, the quantization noise in the frequency domain to adjust is to "shape" the noise below Use of scale factors. To that end, like it is known the spectrum into several groups of spectral coefficients divided, which are called scale factor bands, to which a single scale factor is assigned. On Scale factor represents a multiplication value that uses is the amplitude of all spectral coefficients in to change this scale factor band. That mechanism is used to map the quantization noise in the spectral range that is generated by the quantizer, so that the energy in each scale factor band of quantization noise under the psychoacoustic Masking threshold in this scale factor band. It is it can be seen that neither quantization nor entropy coding Are processes that favor a constant bit rate. It should be noted that - on the contrary - both Process favor a variable bit rate. For transmission applications however, it is often required that the Encoder has a constant bit rate at the output. To one Delivering a constant bit rate is usually a so-called Bit reservoir used. If the audio signal is like this is that briefly fewer bits than by the external bit rate specified at the output of the encoder is required bits are assigned to the bit reservoir in order to In the case of an audio signal section that has more bits for coding needed to be able to give more bits, so that Bit reservoir is emptied again.

It should be noted that a boundary condition of such Encoder as mentioned the constant output bit rate and that the other constraint is that the Quantization noise less than or equal to the psychoacoustic Masking threshold is to allow it through the audio signal is masked or covered.

In the following, possibilities are discussed, what about is done when the "inner bit rate" of the encoder of deviates from the outer constant output bit rate. Is the inner one Bit rate so low that, for example, the bit reservoir is filled to its maximum value, so exists of course not a problem as the quantizer can then be controlled so that it is now even finer quantized than necessary, causing more bits to quantize are needed. This will continue until the "outer" constant bit rate is reached.

More critical, however, is the case where the "inner bit rate" of the encoder is higher than that required on the output side constant bit rate. This case will occur if that Audio signal is difficult to encode, i. H. if the encoder have to spend many bits to encode the audio signal, which is also referred to as the "high load" of the encoder can be. For the transformation coding exists the notion that it encodes tonal pieces relatively efficiently can, however, be noisy signals that are relative have high energies, and also a relative one have a complex spectrum, such as language or drum or drum music, relatively little compressed can be. Also signals that are transient, i. H. who have an irregular timing can only be relative be encoded at great expense if there are no coding artifacts should be generated. In the case of transient signals already when windowing from long windows to shorter ones Window switched to better temporal resolution too reach, or to achieve that the quantization noise only over a smaller number of audio samples "Smeared". In the case of short windows fall much more page information.

An encoder that determines that the output bit rate is not is enough, and which has already "emptied" the bit reservoir has now several options to its internal bit rate "violent" to reduce the criterion of constant Output bit rate to meet. One way is to avoid switching to short windows. This however leads to audible coding artifacts.

Another option is the psychoacoustic Deliberately violating the masking threshold during quantization, to quantize coarser than is actually necessary, to achieve a lower bit rate. This also leads to audible interference.

Another option is the audio bandwidth to decrease, d. H. no longer the full audio bandwidth encode but from a certain of the output bit rate dependent spectral frequency the overlying spectral values to be set to 0 in order to reduce the output bit rate. This method does not lead to audible quantization disturbances, however, leads to a loss of highs in the audio signal. However, this loss is often perceived less strongly as an audible quantization noise.

A particular problem when encoding stereo signals consists of the effect called "stereo unmasking", which is briefly outlined below. Becomes a normal L / R coding is used, so both the left channel and the right channel itself also transformed, quantized and encoded so that in the left channel and right channel introduced quantization noise independently for data reduction from the other channel. That means that Left channel quantization noise and quantization noise are not correlated in the right channel. Becomes considers the case that the left and right channels are relative are similar, it means that a listener after decoding will perceive this signal so that for example, there is a speaker in the middle. The "stereo unmasking" effect is that because of the fact that the quantization noise in the two channels is not correlated, the quantization noise of the left Left channel and the quantization noise of the right channel is perceived on the right. A high masking of the noise but only takes place in the middle, where that too Useful signal is, but not left and right.

The M / S coding therefore has, besides its data rate reducing Effect with special signals also the advantage that the quantization noise in both the left channel as well in the right channel with the quantization noise of the other channel is correlated, so that the quantization noise takes place in the middle and there of that Useful signal essentially completely or much better than is covered up in the uncorrelated case. It is different Case where the left and right channels are relatively dissimilar are. If M / S coding is used here, then due to the stereo effect, the useful signal is either on the left or be right, while due to the M / S coding the Quantization noise is correlated and more in the middle lies. Stereo unmasking also takes place here, so to speak instead of.

Recently, more and more scalable audio encoders are becoming examined. Scalable audio encoders are arranged that its output bitstream has at least a first one and has a second scaling layer. A decoder, which is simply designed, is made from the scaled bit stream just take the first scaling layer, for example an encoded audio signal with reduced bandwidth has or a with a simple coding algorithm encoded audio signal. Another decoder, the is fully designed, both the first scaling layer as well as take the second scaling layer out of the bit stream, around the first scaling layer with a first decoder to decode, and then the second scaling layer also decode that alone or together an audio signal with the decoded first scaling layer with full bandwidth.

Scalable encoders are particularly useful in the area of stereo signals desirable, because here as the first scaling layer Mono signal, i.e. H. the middle channel, can be used while as the second scaling layer z. B. the side channel can be taken. A simple decoder or a decoder, which is designed for fast operation is only deliver the mono signal while a better decoder or a decoder, in which the speed of the Transfer is not the most crucial criterion, besides take the side layer from the mono or middle layer to a full stereo signal at the output of the decoder to create.

Various structures exist for the construction of the scaling layers Possibilities. The first scaling layer can from the second scaling layer or from any one Number of further scaling layers in the audio coding process itself, in the audio bandwidth, in the audio quality, regarding Mono / Stereo and or a combination of the above Distinguish quality criteria or other conceivable criteria. For high coding efficiency, the aim is that the second scaling layer has the smallest possible number has bits, or that a decoder that the decoded second scaling layer, as extensively as possible also uses the first scaling layer. If a scalable Encoder for stereo signals is considered, the as the first scaling layer, the middle signal, i. H. the Mono signal, and the second layer is the side channel supplies, it can be seen that its overall efficiency The better the M / S coding is used, the better. However, this requirement applies to certain stereo signals contrary to bit efficiency, namely with stereo signals, that have high stereo channel separation. On the other hand, delivers the M / S processing a certain "natural" scalability and leads to a correlation of the quantization noise in the left channel and in the right channel.

The M / S coding problems mentioned apply the more, the more an audio signal to be encoded has its properties regarding the M / S coding suddenly changes.

Suddenly an audio signal to be encoded no longer has the Property that the left channel is similar to the right, the M / S coding gain is eliminated. One consequence is therefore in usually an increase in the quantization disturbance beyond the psychoacoustic hearing threshold and / or one Reduction of the audio bandwidth depending on the specific implementation of the encoder.

This problem is particularly severe, but not only noticeable in scalable audio coding, and in particular where the so-called mono stereo scalability is used as stated above.

The object of the present invention is a Device and method for processing a stereo audio signal to create that to less audible interference leads.

This task is accomplished by a processing device a stereo audio signal according to claim 1 and by a method of processing a stereo audio signal after Claim 18 solved.

The present invention is based on the finding that that it is often cheaper for stereo audio signals a high stereo channel separation to avoid a higher one Audio bandwidth and / or less audible interference in comparison to achieve the case where the stereo channel separation is maintained while the audio bandwidth is reduced or interference introduced by quantization become audible.

Experience has shown that a listener becomes audible quantization disturbances feel more uncomfortable than less Stereo channel separation. Audible quantization disorders are generally a foreign body in an audio signal while a Receiver of a stereo signal processed according to the invention does not necessarily know how the stereo channel separation of the Output signal was and thus a lower stereo channel separation is not perceived as a coding artifact.

A reduction in stereo channel separation is thus used the encoder bit rate in general decrease, or reduce to a predetermined value.

A device according to the invention for processing a stereo signal, that a first channel and a second channel comprises a device for analyzing the stereo audio signal, to get a measure of a lot of bits which is needed by an encoder to get the stereo audio signal to encode using an encoding algorithm and means for modifying the first and of the second channel to a modified first and one to obtain modified second channel, the device to modify to the device for analysis responds to be effective when the measure of the amount of Bits exceed a predetermined amount, and being set up is designed to modify such that a Sum signal from the first and second modified channel at least according to a characteristic of the signal that is similar to the energy of the signal changes, essentially equal to the characteristic of a sum signal from the first and second channel, and that a difference signal from the first and second modified channels compared to Difference signal attenuated from the first and second channels is.

It should be noted that the characteristic is similar runs to energy, which can be energy itself, but also z. B. the sum of squared samples in a certain period of time, the sum of squared spectral values in a certain frequency range, the sum of sample amounts in a certain period of time or the sum of squared spectral values in a given Frequency range is or a combination between two or more of the above characteristics. For simplicity will be an example in the following of energy as a characteristic that is similar to energy runs, spoken.

Modifying the stereo audio signal, i.e. H. reducing the channel separation, is carried out on the condition that the volume of the signal does not fluctuate. A reduced channel separation itself does not become annoying artifacts result in a fluctuation in the decoded signal However, volume. Therefore, the first and the second channel, e.g. B. the left channel and the right channel, modified such that the volume, i.e. H. the sum signal, compared to the unmodified first and second channels at least in terms of energy and preferably even in terms of signal remains essentially the same while the difference signal is subdued.

The preprocessing of the stereo signal according to the invention is always use when it is determined that the quantity bits needed to encode the stereo audio signal gets too high. The measure of the amount of bits used for Coding of the stereo audio signal may be needed the stereo audio signal by analyzing it for various Ways are derived.

First, the center and side channels of the stereo audio signal to be considered due to an energy ratio or a difference in the logarithms of the Energies to determine how many bits are needed. Without having to determine the exact number of bits, it is reasonable to conclude that in the case of a small Energy ratio between the center and side channel, d. H. in the case of channels of approximately the same size, a high number of bits will be needed. The lower the energy ratio between the middle and the side channel, so more attenuation of the side channel will be necessary to get one to achieve certain output bit rate. A small energy ratio lies between the middle and the side channel before when the original audio signal has a high stereo channel separation has, for example if the left channel a lot Has energy while the right channel is essentially noise Has. However, there is also a small energy ratio before if the language of a speaker is in the left channel, and if in the right channel the language of another speaker is what causes the left channel and the right Channel may have the same amount of energy that however, both channels are uncorrelated. In this case, too there is a high stereo signal separation and the center channel and the side channel will be a relatively small difference have the logarithms of energy.

Another way to determine the measure for a However, the amount of bits is independent of the nature the middle channel and the side channel in it To look at the encoder itself. A measure of that of one The number of bits required by the encoder is the so-called perceptual Entropy (PE), which is equal to the energy ratio between the useful audio signal and that calculated for the useful audio signal psychoacoustic listening threshold is. If the PE is big, can be concluded that the audio signal is a relative has low concealability. On the other hand, if the PE is small, d. H. the energy of the useful signal is only slightly above that psychoacoustic listening threshold, so the useful signal only be quantized relatively roughly, and the quantization noise is still below the psychoacoustic listening threshold "hidden". It is found that the sum of the PE, preferably averaged over a certain time of the left channel and that, also preferably over one averaged over time, PE for the right channel over is a predetermined value, the Side channel attenuated to the required number of bits to reduce. This alternative aspect of the present Invention is therefore not concerned with the individual Appearance of the center and side channels, but with the Stereo audio signal itself, which is not regarding its M / S coding ability is assessed, but its general Audio coding capability, d. H. the difficulty of doing the same encode to achieve a specific target bit rate.

A generalization of the second aspect is some other size as a measure of the amount of bits too use, which indicates the "load" of the encoder. Such a variable can also be a signal, for example, which indicates due to transient properties of the audio signal, that an audio encoder use short windows for windows must, since the fact is that short windows are not most recently due to the increased number of page information require a higher bit rate. In order to The present invention can thus cover the entire range of Control variables of an audio encoder are used to get a Measure to find that or how much the side channel is damped must be set to the encoder output bit rate reduce.

Preferred embodiments of the present invention lead an increasing or decreasing in time Attenuation of the side channel through to prevent a Listener immediately perceives the decreasing stereo channel separation, but that the reduction in stereo channel separation gradually occurs or the stereo channel separation increases gradually increases to the encoder side To mask manipulation of the stereo audio signal as well as possible.

It should be noted that to maintain a non-fluctuating Volume due to modifying the sum signal of the modified left and right channels necessarily to the sum signal of the unmodified left and right channel must be identical, but that it is sufficient that only the energies of the two sum signals in are substantially the same or in a predetermined ratio lie to each other. A listener doesn't know how big was the volume of the unmodified stereo audio signal and therefore will not perceive it as a disturbance if by the preprocessing a volume change in the direction higher volume or lower volume introduced has been. Because of the ease of implementation however, it preferred that this ratio be 1.

Fig. 1

ein Prinzipblockschaltbild der erfindungsgemäßen Vorrichtung zum Verarbeiten eines Stereoaudiosignals;

Fig. 2

eine detailliertere Darstellung einer bevorzugten Ausgestaltung der Einrichtung zum Modifizieren; und

Fig. 3

ein Blockschaltbild einer erfindungsgemäßen Vorrichtung als Vorverarbeitungsstufe für einen skalierbaren Codierer mit Mono/Stereo-Skalierbarkeit.

Preferred embodiments of the present invention are explained in detail below with reference to the accompanying drawings. Show it:

Fig. 1

a basic block diagram of the inventive device for processing a stereo audio signal;

Fig. 2

a more detailed representation of a preferred embodiment of the device for modification; and

Fig. 3

a block diagram of a device according to the invention as a preprocessing stage for a scalable encoder with mono / stereo scalability.

Fig. 1 shows a block diagram of the device according to the invention to process a stereo audio signal that is on an input 10 is fed into the device and has a first channel L and a second channel R. The Stereo audio signal in the form of the first channel L and the second Channel R is on the one hand in a device 12 for analysis of the stereo audio signal, and on the other hand also in a device 14 for modifying the first and second channels fed to at an output 16 shows a modified first channel L 'and a modified one to obtain second channel R '. Generally will the modified first channel L 'and the modified second Channel R 'at output 16 from the unmodified first channel L and from the unmodified second channel R at input 10 differ in that the applied to output 16 modified Stereo audio signal has a lower channel separation than have the unmodified stereo audio signal at input 10 becomes.

The device 12 for analyzing the stereo audio signal determines a measure of an amount of bits that are in one Fig. 1 encoder, not shown, is required to the stereo audio signal using one by the encoder to code the given coding algorithm. The measure of that Bit amount is determined by the device 12 for analysis a signal path 18 of the device 14 for modification fed. Exceeds that supplied via signal path 18 Measure for the amount of bits a predetermined measure, so the device 14 effective for modification to the first channel L and to modify the second channel R. According to the invention the modification of the first and second channels becomes like this performed that the energy of the sum of the modified Stereo audio signal at output 16 in a predetermined ratio and preferably substantially equal to the energy of the unmodified stereo audio signal at input 10 is, however, the difference signal, which apart from the factor of z. B. 0.5 corresponds to the side channel, in modified stereo audio signal at output 16 compared to unmodified stereo audio signal attenuated at input 10 is.

In Fig. 1 are two ways of feeding the device 12 shown for analysis, individually or in Combination can be used. The first option is represented by a left arrow 15a, so to speak represents a feed forward, d. H. the establishment to analyze the stereo audio signal is not with the modified signal L, R fed. The other possibility consists of the device 12 for analysis with the modified signal L ', R' to feed. Especially in cases in which the attenuation of the side signal is slow in time expires, it is irrelevant whether the damping is dependent from the current unmodified signal or from a the last processing blocks of the modified signal is controlled to a certain extent in terms of feedback. So that is it doesn't matter whether the stereo audio signal itself is direct is analyzed, or indirectly based on a previous one modified signal.

In the following, various configurations of the Means 12 for analyzing the unmodified stereo audio signal received at entrance 10. A possibility is that the means 12 for analyzing both the center and side channels of the stereo audio signal forms and then the relationship of the energies of the middle and of the side channel viewed. The energy ratio between the middle and the side channel is preferably over a certain time averaged, for example in the order of magnitude of 10 audio frames, which is a value of Corresponds to 200 ms if an MPEG-2 AAC encoder is used as the audio encoder is used, which has a frame length of about 20 ms can have. Regarding the MPEG-2 AAC encoder is on the standard ISO / IEC 13818-7, in which the individual Function blocks of an audio encoder and an audio decoder and their interaction is described in detail are.

It is found that the energy ratio or the difference the logarithms smaller than a certain one depending on the Use case is the value to be determined empirically, for example can be chosen to 6 dB, the facility 14 activated for modification in order to dampen the Side channel as it is referring to FIG. 2 will be run in more detail. According to the im Outlined first aspect of the present Invention works the device 12 for analyzing the Stereo audio signal thus based on a direct investigation the MS coding capability of the stereo audio signal. at an implementation of this first aspect of the present Invention is the inventive device for processing of the stereo audio signal only then the side channel attenuate when the signal is no longer so good MS coding ability because, for example, both channels either energy and / or signal dissimilar to each other are. According to this aspect, stereo channel separation becomes whenever reduced to keeping the original Stereo channel separation at too high an output bit rate would result, and if stereo channel separation at all was high.

According to another aspect of the present invention the attenuation of the side channel to reduce the output side Encoder bit rate used regardless of whether the stereo audio signal has a certain MS coding capability or not. This second aspect of the invention proceeds from this from that even in the case of a small stereo channel separation still further attenuation of the side channel achieved can be set to a predetermined output bit rate of the audio encoder not to exceed. This is independent of the MS coding capability of the audio signal the number of bits estimated that is required to encode the audio signal.

As is known in the art, modern audio encoders use and for example an MPEG-2 AAC audio encoder, a psychoacoustic model that is used by an audio signal to be coded, the frequency-dependent psychoacoustic Calculate masking threshold. Roughly said provides the psychoacoustic model as a psychoacoustic Masking threshold an energy value for each scale factor band. Is the quantization noise introduced by the quantizer below the energy value or is that through the quantization noise introduced is equal to that Energy value, so according to the psychoacoustic theory the quantization noise introduced is essentially inaudible his.

The energy ratio or the difference of the logarithms of the Audio signal itself and its psychoacoustic masking threshold, also called Perceptual Entropy (PE) provides a measure of how many bits are to be encoded of the audio signal are required. If the PE is high, so many bits are required because the masking ability of the Audio signal is relatively low and therefore finely quantized must become. On the other hand, if the PE is small, relatively few will be Bits because the audio signal is masked relatively well, and therefore only a relatively rough quantization is required is.

According to a preferred embodiment, the second Aspect of the present invention the measure of the amount of bits determined as follows. The PE values for the individual Scale factor bands are integrated over frequency, d. H. summed up. This will be for both the left performed for right channel as well. Then will the PE sum for the left channel to the PE sum for the right channel summed. This sum PE value from left and the right channel represents the bit requirement for a frame Sum channel PE value is then preferably still above a certain number of frames, such as B. 10, averaged, by an averaged PE value for the stereo audio signal receive. If this averaged PE value is greater than or equal to a predetermined one that is typically to be determined empirically Value, the multiplication facility is activated, to dampen the side channel.

Generalized as a measure of the amount of bits, any encoder will need, any other control variable be used, which is a measure of the "load" of the encoder represents such. B. a control signal of the encoder, which signals the use of short windows in windows. Windows with short windows lead to one per se higher number of bits because shorter windows are not so bit-saving can be coded like longer windows.

Regarding the amount of attenuation of the side channel exist several options here regarding their effort differ. The easiest way is in agreeing a predetermined damping value as the target value, which, for example, should be determined empirically can. However, another option is to to adaptively determine the damping value, d. H. the side channel to dampen a predetermined increment amount, and then again to see if the number of bits is already there has decreased enough or not. It can then turn into a new iteration loop with a further increment damping amount be boarded to determine again whether the number of bits is already sufficiently small. This The procedure can be repeated until the encoder required number of bits in a target corridor lies. However, it can be seen that the computing time and Implementation effort in the case of adaptive damping adjustment much higher than in the case of a given one Damping is. On the other hand, an adaptive damping adjustment delivers the best and most accurate results.

2, in which one detailed representation of the device 14 for modification according to a preferred embodiment of the present Invention is shown. The means 14 for modifying can be interpreted as having a first input 20a for the first channel L and a second input 20b for the second channel R. The device 14 comprises a first multiplier 22a for multiplying the first channel L with a certain factor x, a second Multiplier 22b for multiplying the first channel L by a factor y, a third multiplier for multiplying of the second channel R with the factor x and finally a fourth multiplier 22d for multiplying the second channel R with the factor y. In addition, the Means 14 for modifying a first summer 24a for summing the output signal of the first multiplier 22a with the output signal of the fourth multiplier 22d, and a second summer 24b for summing the output signal of the second multiplier 22b with the output signal the third multiplier 22c. At exit 26a of the first Summierers 24a is the modified first Channel L 'on, and at the output 26b of the second adder 24b is the modified second channel R '.

In the following, the determination of the two multiplication factors x, y is dealt with in order to achieve a damped side channel, while the center channel at the output 26a, 26b is equal to the center channel at the input 20a, 20b of the device 14, which is shown in FIG 2 is shown. The following matrix applies to the signal processing carried out by the modification device 14: L '= xL + yR R '= yL + xR

The task now is to determine x and y, so that: L '+ R' = L + R = 2 M = 2 M ', and that applies: L'- R '= S' = damping * S = damping * (LR)

The result is: M = 0.5 (x + y) (L + R)

Since M should not be modified by the processing, the following equation also applies: x + y = 1

For the side channel: S = 0.5 (x - y) (L - R)

From equation (7) it follows that S reduces by the factor x-y is, or, expressed logarithmically, by 10 · log10 (x-y) dB = att, is attenuated. att stands for the Attenuation, and it is true that att is less than 0 dB.

For attenuation in dB steps: att (in dB) = 20 * log10 (xy)

The following expression then results from equation (8): exp (0.05 att) = x - y

Equation (6) and equation (9) result in equation (10) for x and equation (11) for y. x = 0.5 * (1 + exp (0.05 att)) y = 0.5 * (1 - exp (0.05 att))

The attenuation "att" (in dB) is dependent on one of the control variables described. This results in with equations (9) and (10) the factors x and y for the damping matrix represented by FIG is equally reflected in equations (1) and (2). In order to save implementation effort and computing effort, does not have to be a fully adaptive damping adjustment att can be done, but it can be a specific one Attenuation value att, which has been determined empirically, is used if the measure of the amount of bits is one exceeds the predetermined limit.

According to the invention, the damping is not increased suddenly, because a decrease in channel separation that occurs suddenly goes to an audible disturbance or could surprise the listener, for example if a speaker was initially placed on the left and up is perceived once in the middle. Therefore, in the case where which is determined to dampen the side channel, a gradual damping of the side channel, for example using a predetermined increment value, such that vividly speaking the news anchor slowly "wanders" from the left to the center. In the opposite case, it is found that the Measure of the amount of bits again smaller than the predetermined The damping is not abruptly canceled, but slowly returned to 0, so that um in the example, the speaker slowly stays away from the Center to the side "wanders". This gradual damping or gradual cancellation of damping should be as slow as possible take place so the damping of the side channel is practical is not perceived. The reduction in damping must however take place so quickly that the encoder due to the high bit rate at the output does not start the psychoacoustic Violate masking threshold or audio bandwidth to remove. According to the invention, encoders that have a bit reservoir mechanism, this bit reservoir exploited to slowly increase the damping until the Target value has been reached at which the damping is so high that the predetermined bit rate at the output of the encoder is observed can be. If the damping is then canceled again the bit reservoir can be emptied again.

In the implementation shown in Figure 2, one was Obtain boundary conditions for determining x and y in such a way that the sum signal that goes down to the middle channel Factor corresponds to 0.5, was not changed. However, there are Signals conceivable where the left and the right channel are similar, but a phase shift in the range of 180 degrees to each other. It should be noted that such signals are not particularly common because they are not well presented with mono playback devices can be. Nevertheless, such signals are but conceivable. In this case, the center channel M would be small and the side channel get big. Then S would be so strong that S becomes smaller than M would also be damped Overall volume can be greatly influenced. In contrast to however, it is for a reduction in stereo channel separation a listener intolerable if the volume is high fluctuates, regardless of the audio signal itself. A listener will find such a disturbance annoying.

To avoid this problem, it is preferred additionally in the device 12 for analyzing, whether the phase shift of L and R is close of 180 degrees. If this is found, it can be done easily the sign of R can be reversed. Then it goes originally wanted spatial stereo effect lost, however the effect of reduced volume is avoided what will bother a listener less.

As an alternative to the sign reversal, the M channel could also be in the device for modification or in a downstream Encoder level increased to a certain value such that the energy of the modified M-channel in a predetermined ratio to the energy of the M-channel of the unmodified stereo audio signal. For the energy ratio a value of 1 is preferred, but also a certain gain by the modifier means or damping can be performed, but always that Relationship to the unmodified stereo audio signal essentially should be maintained so that a listener does not significant volume fluctuations due to preprocessing will perceive. Of course, small ones Volume fluctuations are not so problematic and sometimes even imperceptible. Large volume fluctuations however, a test listener will find it annoying.

At this point it should be noted that it is irrelevant is whether at the input 10 of the inventive device for Processing a stereo audio signal discrete-time samples or spectral values. All operations to analyze the stereo audio signal can both with discrete-time samples as well as spectral values be performed. In addition, all operations in the device for modification with both time-discrete Samples as well as spectral values performed become. The processing device according to the invention of a stereo audio signal could therefore also be Time-frequency transformation level of a time / frequency transformation-based Be arranged encoder, such as. B. an MPEG audio encoder. This concept even results the additional possibility of stereo preprocessing can be made frequency-selective, d. H. that for example a different attenuation of the signal S in Depending on the frequency can be performed. This is particularly useful because of the possibility of directional location of human hearing not for all frequencies is equally sensitive. Thus becomes the invention Processing carried out by spectral value, so can spectral values of the side channel can be damped the more less human hearing in this frequency range hears directionally, while spectral values do not or only be touched very little, which are in frequency ranges, in which human hearing is a good directional fix supplies.

It should be noted that in modern audio encoders anyway in terms of frequency using the so-called M / S mask it is determined where an M / S coding is carried out and where L / R coding is better. In this The processing according to the invention would only fall on the frequency ranges are applied in which an MS coding is present, d. H. in which the MS mask is set is. Alternatively, the MS mask could be used in more bands be set, d. H. MS coding can be performed in this in comparison to the known method additional MS bands the side channel is attenuated to Comply with bit rate requirements.

In the following, reference is made to FIG. 3, in which a Device for processing a stereo audio signal shown which is in addition to the puncture blocks shown in FIG also an MS encoder 30 and a scalable Encoder 32 includes a scaled output Bitstream BS outputs. The MS encoder 30 includes how it is known in the art to have a summer 30a for summing of the modified left channel L 'and the modified right channel R 'to multiply by a multiplier 30b to which a factor of e.g. B. 0.5 is assigned to generate the multiplied center channel. In addition, the MS encoder 30 includes a subtractor 30c and a further multiplier 30d to the modified To generate side channel S 'which is opposite a side signal, that from the unmodified stereo audio signal is formed at the input 10, is damped. The middle channel M ' and the side channel S 'are both scalable Encoder 32 is fed, which preferably has a mono stereo scalability having. The first scaling layer is represent the mono signal M ', and the second scaling layer will include the modified side channel S '. Further Scaling options, such as B. that the modified or unmodified mono channel M 'additionally band limited and that in the second scaling layer next to the modified side channel also contain the upper mono band is possible.

The effect of scalability on the mono stereo encoder 32 is particularly cheap, if not LR coding but MS coding is used. The invention Stereo signal processing by devices 12 and 14 is therefore especially in connection with the scalable encoder 32 particularly advantageous. A mono stereo scalability to obtain, namely, MS coding can be used if they are actually compared to the LR coding is no longer preferable. This is what it does achieved that the side channel at the entrance of the scalable Encoder 32 damped compared to the unmodified case is.

3 is a dashed signal path 36 from scalable encoder 32 for device 12 for analysis located. This dashed signal path 36 is intended to symbolize that certain measures to measure the amount to derive bits required by the scalable encoder to encode the stereo audio signal at input 10, do not have to be calculated directly in the device 12, but from the scalable encoder into the device 12 can be output, such as B. Perceptual Entropy PE, the reference to the use of short windows, etc. That means that these function blocks are not both in the facility 12 for analysis as well as in the scalable encoder 32 must be present, but that their implementation only in the scalable encoder 32 is sufficient.

In this case, the means for modifying 14 would to determine the measure 18 for the amount of bits, initially none Carry out modification. The device shown in Fig. 3 would be in a "pre-run mode" where no bit stream is written, but where only the required degree of damping for the side channel determined becomes. In the subsequent coding mode, in which the bit stream BS then written by the scalable encoder the means 14 for modifying with accordingly fixed factors x, y work.

3 shows the device with spectral values operated for the first channel L and the second channel R, and the scalable encoder is a time / frequency transform encoder, so would the level of the scalable encoder 32, which performs the time-frequency transformation, be connected upstream of input 10. The facilities 12, 14 and 30 would then be embedded in the scalable encoder 32.

The signal paths 36a, 36b illustrate that the modified ones Channels without M / S coding to a scalable encoder can be directed so that it can then determine can determine whether M / S or L / R coding is cheaper.

Claims (18)

1. Device for processing a stereo audio signal having a first channel (L) and a second channel (R), comprising:

   a means (12) for analyzing said stereo audio signal or a signal derived from said stereo audio signal to obtain a measure for the quantity of bits required by a coder (32) to code said stereo audio signal using a coding algorithm; and

   a means (14) for modifying said first and said second channel (L, R) to obtain a modified first and a modified second channel (L', R'),

   said means (14) for modifying responding to said means (12) for analyzing to become effective if said measure (18) for the quantity of bits exceeds a predetermined measure, and

   said means (14) for modifying being designed in such a way that a characteristic, having a similar course as the energy of the sum signal, of a sum signal of said first and said second modified channel (L', R') is in a predetermined relation to the characteristic of a sum signal of said first and said second channel (L, R) and that a difference signal of said first and said second modified channel (L', R') is attenuated in contrast to a difference signal of said first and said second channel (L, R).
2. Device according to claim 1, in which said means (12) for analyzing comprises:

   a means for determining the characteristic of the sum of said first and said second channel over a predetermined time period;

   a means for determining the characteristic of the difference of said first and said second channel over a predetermined time period; and

   a means for forming the relation of the characteristic of the sum of said first and said second channel and the characteristic of the difference of said first and said second channel, the relation of the characteristics being the measure (18) for the quantity of bits.
3. Device according to claim 1, in which said means (12) for analyzing comprises:

   a first means for determining a first characteristics relation between said first channel and said psychoacoustic masking threshold of said first channel over a predetermined time;

   a second means for determining a second characteristics relation between said second channel and said psychoacoustic masking threshold of said second channel over a predetermined time; and

   a means for summing said first and said second characteristics relation, the sum of said first and said second characteristics relation hinting to said measure (18) for the quantity of bits.
4. Device according to claim 1, in which said coder (32) is arranged to use, responsive to the temporal structure of said stereo audio signal, long or short windows for transforming said temporal stereo audio signal into a spectral stereo audio signal, and in which said means (12) for analyzing is arranged to detect whether short or long windows are used in said coder (32), said measure for the quantity of bits being that short windows are used.
5. Device according to one of the preceding claims, in which said means (14) for modifying is arranged to become effective in such a way that the difference signal of said first and said second channel is gradually attenuated departing from no attenuation to a certain attenuation and to be effective in such a way that the attenuation is gradually reduced from the determined attenuation to no attenuation.
6. Device according to claim 5, in which the speed of the attenuation is selected to be as slow as possible, but not that fast that a bit reservoir mechanism of said coder (32) is used that said coder (32) does not reduce the audio bandwidth nor impedes a psychoacoustic masking threshold when quantizing.
7. Device according to one of the preceding claims, in which said means (14) for modifying is arranged to adaptively attenuate the difference signal depending on the determined measure.
8. Device according to claim 2, in which said means (14) for modifying is arranged to attenuate the difference signal depending on a characteristics relation generated by said means for building the characteristics relation so that the attenuation of the difference signal is high when the characteristics relation is small and that the attenuation of the difference signal is low when the characteristics relation is high.
9. Device according to claims 7 or 8, in which said means (14) for modifying is designed in such a way that it adaptively attenuates the difference signal in such a way that the characteristics relation of the difference signal to the sum signal is essentially equal to a predetermined value.
10. Device according to one of the preceding claims, in which said means (14) for modifying comprises:

    a first multiplier (22a) for multiplying said first channel (L) by a first factor (x);

    a second multiplier (22b) for multiplying said first channel (L) by a second factor (y);

    a third multiplier (22c) for multiplying said second channel (R) by said first factor (x);

    a fourth multiplier (22d) for multiplying said second channel (R) by said second factor (y);

    a first summer (24a) for summing the output signal of said first multiplier (22a) and the output signal of said fourth multiplier (22d) to generate the modified first channel (L'); and

    a second summer (24b) for summing the output signal of said third multiplier (22c) and the output signal of said second multiplier (22b) to generate the modified second channel (R'),

    said first and said second factor (x, y) being selected in such a way that the sum signal of said first and said second channel and the sum signal of said modified first and second channels are essentially equal and that the difference signal is attenuated by a certain factor.
11. Device according to one of the preceding claims, in which said means (12) for analyzing further comprises:

    a means for determining whether a phase angle between said first and said second channel (L, R) has a value in the vicinity of 180°; and

    said means for modifying (14) further comprising:

    a means for reversing the sign of a channel (L, R) if the phase angle is in the vicinity of 180°.
12. Device according to one of the preceding claims, in which said first and said second channel (L, R) of said stereo signal are given by spectral values having been generated from a temporal stereo signal by a transfer into the spectral range, said means for modifying (14) being arranged to perform a frequency-selective attenuation of said difference signal.
13. Device according to claim 12, in which said means for modifying is arranged to more strongly attenuate in a frequency range in which the direction finding of the human hearing is reduced than in a frequency range in which the direction finding of the human hearing is not reduced.
14. Device according to one of the preceding claims, further comprising:

    a center/side means (30) for generating a center channel (M') equaling half the sum of said modified left (L') and said modified right channel (R');

    a side means (30) for generating a side channel equaling half the difference of said modified first channel (L') and said modified second channel (R'); and

    a scalable coder (32) arranged to code said center channel (M') and to write into a bit stream (BS) as a first scaling layer, and further being arranged to code said side channel (S') and to write into said bit stream (BS) as a second scaling layer.
15. Device according to claim 14, in which said scalable coder (32) is arranged to use a bit reservoir means for the case in which the measure for the quantity of the bits exceeds a predetermined value, in order not to reduce the audio bandwidth and/or not to impede the psychoacoustic masking threshold.
16. Device according to one of the preceding claims, in which the characteristic having a similar course as the energy is the energy itself, the sum of squared sample values in a certain time period, the sum of squared spectral values in a certain frequency range, the sum of sample value amounts in a certain time period and/or the sum of squared spectral values in a certain frequency range.
17. Device according to one of the preceding claims, in which said stereo audio signal is processed block-wise, and in which said signal used in analyzing and derived from said stereo audio signal is the modified signal of a preceding processing block.
18. Method for processing a stereo audio signal having a first channel (L) and a second channel (R), comprising:

    analyzing (12) said stereo audio signal or a signal derived from said stereo audio signal to obtain a measure for the quantity of bits required by a coding algorithm to code said stereo audio signal; and

    modifying (14) said first and said second channel (L, R) to obtain a modified first and a modified second channel (L', R') if, in the step of analyzing, a measure (18) for the quantity of bits is determined, which exceeds a predetermined measure, said modifying being performed in such a way that a characteristic, having a similar course as the energy of the sum signal, of a sum signal of said first and said second modified channel (L', R') is in a predetermined relation to a characteristic of a sum signal of said first and said second channel (L, R) and that a difference signal of said first and said second modified channel (L', R') is attenuated in contrast to a difference signal of said first and said second channel (L, R).

Images