MXPA06009110A - Method and device for quantizing a data signal - Google Patents

Abstract

Disclosed is a method for quantizing a data signal of a sequence of data values, comprising frequency-selective filtering of the sequence of data values so as to obtain a sequence of filtered data values, and quantizing the filtered data values in order to obtain a sequence of quantized data values by means of a quantizing stage function which copies the filtered data values onto the quantized data values and the curve of which is steeper below a threshold data value than it is above said threshold data value.

Description

Redundancy is not done based on a single transform, but on two separate transforms. The principle will be discussed subsequently with reference to Figures 12 and 13. The coding begins with an audio signal 902 which has already been sampled and thus is already present as a sequence 904 of audio values or sample 906, wherein the Temporal order of the audio values 906, is indicated by an arrow 908. A listening threshold is calculated by means of a psycho-acoustic model for blocks susceptive to audio values 906 characterized by an ascending numbering by "block #". Figure 13 for example shows a diagram where, with respect to the frequency f, the graph plots the spectrum of a signal block of 128 audio values 906 and b plots the masking threshold, as calculated by a psycho-acoustic model, in logarithmic units. The masking threshold indicates, as already mentioned, up to which intensity frequencies remain inaudible to the human ear, ie, all the tones below the masking threshold b. Based on the hearing thresholds calculated for each block, an irrelevance reduction is achieved by controlling a parameterizable filter followed by a quantizer. For a parameterizable filter, a parameterization is calculated such that its frequency response corresponds to the inverse of the magnitude of the masking threshold. This parameterization is indicated in Fig. 12 by x # (i). After filtering the audio values 906, the quantization with constant step sizes is carried out, such as for example a rounding operation to the next integer. The interference or quantization noise caused by this is white noise. On the decoder side, the filtered signal is "re-transformed" again by a parameterizable filter, the transfer function of which is adjusted to the magnitude of the masking threshold itself. Not only the filtered signal is decoded again with this, but the quantization noise on the decoder side, also conforms to the shape of the masking imbral. In order that the quantization noise corresponds to the masking threshold as precisely as possible, an amplification value a, applied to the filtered signal before quantization, is calculated on the encoder side for each set of parameters or each parameterization. In order for the retransformation to be performed on the decoder side, the amplification value a and the parameterization x are transferred to the encoder as lateral information 910, apart from the current main data, ie the filtered quantized audio values 912. For the reduction of redundancy 914, these data, ie the lateral information 910 and the main data 912, are subjected to loss-free compression, ie entropy coding, which is how the coded signal is obtained. The aforementioned article suggests a size of 128 sample values 906 as a block size. This allows a relatively short delay of 8 ms with a sampling rate of 32 kHz. With reference to the detailed implementation, the article also states that, in order to increase the efficiency of the lateral information coding, the lateral information, ie the coefficients x and y, will only be transferred if there are sufficient changes compared with a set of parameters transferred before, that is, if the changes exceed a certain threshold value. Furthermore, it is described that the preferred implementation is performed in such a way that a current parameter set is not directly applied to all the sample values belonging to the respective block, but that a linear interpolation of the filter coefficients x is used. to avoid audible artifacts. In order to perform the linear interpolation of the filter coefficients, a network structure is suggested so that the filter prevents instabilities from occurring. For the case that a signal encoded with a controlled bit rate is desired, the article also suggests multiplication or selective attenuation of the filtered signal adjusted in scale with the time-dependent amplification factor a, by a factor other than one, such that audible interference occurs, but the bit rate may be reduced in places of the audio signal that are complicated to encode. Although the audio coding scheme described in the aforementioned article already reduces the delay time for many applications to a sufficient degree, a problem in the previous scheme is that, due to the requirement of having to transfer the masking threshold or function of filter transfer from the encoder side, subsequently referred to as a pre-filter, the transfer channel is loaded to a relatively high degree even when the filter coefficients will only be transferred when a predetermined threshold is exceeded. Another disadvantage of the above coding scheme is that, due to the fact that the masking threshold or its inverse have to be made available on the decoder side by the set of parameters x to be transferred, a compromise must be made between the bit rate plus possible low or high compression ratio on the one hand and the most accurate possible approximation or parameterization of the masking threshold or its inverse on the other side. Thus, it is unavoidable that the quantization interference adjusted to the masking threshold by the above audio coding scheme exceeds the masking threshold in some frequency ranges and thus results in audible audio interference to the listener. Figure 13, for example, shows the parametrized frequency response of the parametrizable filter on the decoder side by the graph c. As can be seen, there are regions where the filter transfer function on the decoder side, subsequently referred to as the postfilter, exceeds the masking threshold b. The problem is aggravated by the fact that the parameterization is only transferred intermittently with a sufficient change between parameterizations and interpolating between them. An interpolation of the filter coefficients x #, as suggested in the article, only results in audible interferences when the amplification value a # remains constant from node to node or from node to parametrization to parameterization node. Even if the interpolation suggested in the article also applies to the information value lateral to #, ie the amplification value transferred, audible audio artifacts may remain in the audio signal arriving on the decoder side. Another problem with the audio coding scheme and according to Figures 12 and 13, is that the filtered signal may, due to selective frequency filtering, take an un predictable form where, particularly due to a random overlay of many waves individual harmonics, one or more individual audio values of the encoded signal contribute to very high values which in turn result in a more poor compression ratio in the reduction of subsequent redundancy due to its rare occurrence. It is the object of the present invention to provide a method and a device for quantization of an information signal, in such a way that superior data compression of the information signal can be achieved involving only little deterioration in the quality of the information signal original This object is achieved by a method according to claim 12 and a device according to claim 1. The quantization of the invention of an information signal of a sequence of information values includes selective frequency filtering of the sequence of information values, to obtain a sequence of filtered information values and quantify the filtered information values, to obtain a sequence of quantized information values by means of a quantization stage function that maps the filtered information values to the quantized information values and the course of which is more inclined below a threshold information value than the threshold information value. It has been observed that artifacts artificially generated in the resulting filtered information signal, result from the selective filtering of frequency of an audio signal in which the individual information values, due to a random constructive interference of all or many of the harmonics, take values that are significantly higher than the maximum values of the original signal such as , for example more than twice as high. The central idea of the present invention is that to cut the filtered information signal over a convenient threshold, which is in exemplary form twice as high as the largest possible value of the original information signal to be filtered, such that the Artifacts artificially generated by selective frequency filtering, are removed or smoothed from the filtered information signal, after post-filtering hardly results in any deterioration in quality of the post-filtered information signal after quantification, while cutting or Enlarged size of the quantization stage over a convenient threshold offer huge savings in a bit representation of the filtered information signal. According to a preferred embodiment, the information signal is an audio signal in which selective quantization over or under a certain threshold hardly results in an audible decrease in audio quality with a huge simultaneous reduction in the representation of bits. The function of the quantization step may be provided alternately to quantize all the audio values to a higher quantization step on the threshold value, or a quantization stage function having a flatter course on the threshold value or it has a quantization stage size greater than the threshold value, it is used in such a way that artificially generated artifacts are quantified in a coarser form. ** Preferred embodiments of the present invention will be subsequently described with reference to the accompanying drawings, wherein: Figure 1 shows a block circuit diagram of an audio encoder according to an embodiment of the present invention; Figure 2 shows a flow diagram to illustrate the operation mode of the audio encoder of Figure 1 in the data feed; Figure 3 shows a flow diagram to illustrate the mode of operation of the audio encoder of Figure 1 with respect to the evaluation of the input audio signal by a psycho-acoustic model. Figure 4 shows a flow diagram to illustrate the operation mode of the audio encoder of Figure 1 with respect to applying the parameters obtained by the psycho-acoustic model to the input audio signal; Figure 5a shows a schematic diagram to illustrate the input audio signal, the sequence of audio values that constitute it, and the operative steps of Figure 4 in relation to the audio values. Figure 5b shows a schematic diagram to illustrate the configuration of the encoded signal; Figure 6 shows a flow diagram to illustrate the mode of operation of the audio encoder of Figure 1 with respect to the final processing up to the encoded signal; Figure 7a shows a diagram in which one embodiment of a quantization stage function is illustrated; Figure 7b shows a diagram where another embodiment of a quantization stage function is illustrated; Figure 8 shows a block circuit diagram of an audio encoder that is capable of decoding an audio signal encoded by the audio encoder of Figure 1, according to an embodiment of the present invention; Figure 9 shows a flow chart to illustrate the operation mode of the decoder of Figure 8 in the data feed; Figure 10 shows a flowchart to illustrate the mode of operation of the decoder of Figure 8 with respect to damping the pre-decoded filtered and quantized audio data and the processing of the audio blocks without corresponding lateral information; Figure 11 shows a flow chart to illustrate the mode of operation of the decoder of Figure 8 with respect to the current reverse filtering; Figure 12 shows a schematic diagram for illustrating a conventional audio coding scheme having a short delay time; and Figure 13 shows a diagram wherein in exemplary form a spectrum of an audio signal, a hearing threshold and the post-filter transfer function in the decoder are illustrated. Figure 1 shows an audio encoder according to an embodiment of the present invention. The audio encoder, which is generally indicated by 10, includes a data feed 12 where it receives the audio signal to be encoded, which, as will be explained in more detail later with reference to Figure 5a, consists of a sequence of audio values or sample values, and a data output where the encoded signal is output, the information content of which will be discussed in more detail with reference to Figure 5b. The audio encoder 10 of Figure 1 is divided into an irrelevance reduction part 16 and a redundancy reduction part 18. The irrelevance reduction part 16 includes means 20 for determining a hearing threshold, means 22 for calculating a amplification value, means 24 for calculating a parameterization, node comparison means 26, a quantizer 28 and a parameterizable pre-filter 30 and a first-in-first-in-exit buffer (FIFO = first in first out ) of supply 32, a buffer or memory 38 and a multiplier or multiplication means 40. The redundancy reduction part 18 includes a compressor 34 and a bit rate controller 36. The irrelevance reduction part 16 and the part redundancy reduction 18, are connected in series in this order between the data feed 12 and the data output 14. In particular, the data feed 12 is connected to a data feed of means 20, for determining a hearing threshold and a data feed of the feed buffer 32. A data output of the means 20 for determining a hearing threshold is connected to a feed of the means 24 for calculating a parameterization and to a data feed of the means 22 to calculate an amplification value to pass at a determined hearing threshold thereto. The means 22 and 24 calculate an amplification or parameterization value, based on the hearing threshold and are connected to the node comparison means 26 to pass these results thereto. Depending on the result of the comparison, the node comparison means 26, as will be discussed subsequently, pass the results calculated by the means 22 and 24 as a parameter of feeding or parameterization to the parameterizable pre-filter 30. The parameterizable pre-filter 30 is connects between a data output of the feed buffer 32 and a data feed of the buffer 38. The multiplier 40 is connected between a data output of the buffer 38 and the quantizer 28. The quantizer 28 passes the values filtered audio that can be multiplied or adjusted in scale but always quantified, to the redundancy reduction part 18, more precisely to a data feed of the compressor 34. The node comparison means 26 passes information from which the feed parameters passed to the parameterizable pre-filter 30, can be derived elsewhere redundancy reduction 18, more precisely to another data feed of the compressor 3. The bit rate controller is connected to a control power of the multiplier 40 via a control connection, to provide quantized filtered audio values, as received from the pre-filter 30, to be multiplied by the multiplier 40 by a convenient multiplying , as will be discussed in more detail below. The bit rate controller 36 is connected between a data output of the compressor 34 and the data output 14 of the audio encoder 10, in order to determine the multiplicand for the multiplier 40 conveniently. When each audio value passes quantizer 40 for the first time, the multiplying at the beginning is adjusted to a convenient scale adjustment factor, such as, for example, 1. Buffer 38, however continues to store each filtered audio value for giving the bit rate controller 36, as will be described subsequently, a possibility of changing the multiplying by another step of a block of audio values. If this change is not indicated by the speed controller of 36, the buffer 38 can release the memory occupied by this block. After the configuration of the audio encoder of Figure 1 described above, the mode of its operation will be described subsequently with reference to Figures 2 to 7b. As can be seen from Figure 2, the audio signal, when it has reached audio power 12, has already been obtained by sampling audio signal 50 of an analog audio signal. The sampled audio signal is prmed with a predetermined sampling frequency, which is usually between 32 and 48 kHz. Consequently, in the data feed 12, there is an audio signal consisting of a sequence of audio or sample values. Although the coding of the audio signal is not carried out in a block-based manner, as will be apparent from the subsequent description, the audio values in the data feed 12 at the beginning are combined to form blocks of audio in the stage 52. The combination for forming audio blocks is carried out only for the purpose of determining the hearing threshold, as becomes evident from the following description, and is carried out in a step of feeding the means 20 to determine a hearing threshold. In the present embodiment, it is considered in exemplary form that 128 successive audio values each of which are combined to form audio blocks and that the combination is carried out in such a way that, on the one hand, successive audio blocks are not they overlap and on the other hand they are direct neighbors to each other. This will be discussed in an exemplary manner briefly with reference to Figure 5a. Figure 5a at 54 indicates the sequence of sample values, each sample value is illustrated by a rectangle 56. The sample values are numbered for purposes of illustration, where for reasons of clarity in turn only some sample values of the sequence 54 are illustrated. As indicated by square brackets on the sequence 54, 128 successive sample values each combine to form a block according to the present embodiment, wherein the 128 directly successive sample values form the next block. Only as a precautionary measure, it should be noted that the combination to form blocks can also be done in a different way, exemplarily by superposed blocks or spaced blocks and blocks that have another block size, although the block size of 128 in turn it is preferred since it provides good compensation between high audio quality on the one hand and the smallest delay time possible on the other hand. While the audio blocks combined in the means 20 in the step 52 are processed in the means 20 to determine a block hearing threshold per block, the input audio values will be passed in buffer memory 54 in the feed buffer 32, until the parameterizable pre-filter 30 has obtained power parameters from the node comparison means 26 to perform pre-filtering, as shown in FIG. will describe subsequently. As can be seen from Figure 3, the means 20 for determining a hearing threshold begin their processing directly after sufficient audio values have been received in the data feed 12 to form an audio block or to form the next block of audio. audio, which means 20 monitor by an inspection in stage 60. If there is no complete processable audio block, the means 20 will wait. If a complete audio block to be processed is present, the means 20 for determining a hearing threshold will calculate a hearing threshold in step 62 based on a suitable psycho-acoustic model in step 62. To illustrate the hearing threshold, again reference is made to Figure 12 and in particular, to the graph b that has been obtained based on the psycho-acoustic model, in exemplary form with respect to a current audio block with a spectrum a. The masking threshold that is determined in step 62 is a frequency-dependent function that can vary for successive audio blocks and can also vary considerably from audio signal to audio signal, such as, for example, pieces of rock music to classical music. The threshold of hearing indicates for each frequency a threshold value below which the human ear can not perceive interference. In a subsequent step 64, the means 24 and the means 22 calculate from the calculated hearing threshold M (f) (f indicates the frequency) an amplification value to or set of parameters of N parameters x (i) (I = 1 , ..., N). The parameterization x (i) that means 24 compute in step 64 is provided for the parameterizable pre-filter 30 which is incorporated for example into an adaptive filter structure, as used in linear predictive coding (LPC = linear predictive coding) ). For example, s (n), n = 0, ..., 127, are the 128 audio values of the current audio block and s' (n) the 128 filtered filtered audio values, then the filter is incorporated in exemplary form in such a way that the following equation applies: s' (n) - a (n) - T r ¿(n - fc), K is the filter order and afck, k = 1, ..., K, are the filter coefficients, and the index t is to illustrate that the filter coefficients change in successive audio blocks. The means 24 then calculate the parameterization such that the transfer function H (f) of the parameterizable pre-filter 30 is equal to approximately the inverse of the magnitude of the masking threshold M (f), ie in such a way that it applies the following : K (í, t) M ± f where the dependency t in turn is to illustrate that the masking threshold M (f) changes for different audio blocks. When the pre-filter 30 is implemented as the aforementioned adaptive filter, the filter coefficients a will be obtained as follows.- the inverse discrete Fourier transform of | M (f), t) | 2 in the frequency for the block at time t, results in the objective auto-correlation function r ^ d). Then, the ay is obtained by solving the linear equation system: In order that instabilities do not arise between the parameterization in the linear interpolation described in more detail below, a network structure is preferably used for the filter 30, where the filter coefficients for the network structure are re-parameterized, to form reflection coefficients. With regard to more details regarding the pre-filter design, the calculation of the coefficients and the re-parameterization, reference is made to the article by Schuller etc., mentioned in the introduction to the description and in particular to page 381, Division III, which is incorporated herein by reference. While the means 24 consequently calculate a parameterization for the parameterizable pre-filter 30 such that its transfer function is equal to the inverse of the masking threshold, the means 22 calculate an interference power limit based on the hearing threshold, that is, a limit indicating that interference power is allowed to the quantizer 28 to input the audio signal filtered by the pre-filter 30 so that the quantization interference on the decoder side is below the hearing threshold M ( f) or exactly match it after post- or reverse filtering. The means 22 calculates this interference power limit as the area below the square of the hearing threshold magnitude m, ie as | M (f) | 2. The means 22 calculates the amplification value a, from the interference power limit when calculating the root of the fraction of the quantization interference power divided by the interference power limit. The quantization interference is the interference caused by the quantizer 28. The interference caused by the quantizer 28 is as will be described below, white noise, and thus frequency-independent. The quantization interference power is the power of the quantization interference. As has been evident from the above description, the means 22 also calculates the interference power limit apart from the amplification value a. Although it is possible for the node comparison means 26 again to calculate the interference power limit from the amplification value a obtained from the means 22, it is also possible that the means 22 also transmit the interference power limit determined to the comparison means of node 26 apart from the amplification value a. After calculating the amplification value and the parameterization, the node comparison means 26 verify in step 66, whether the newly calculated parameterization differs more than a predetermined threshold from the last current parameterization passed to the parameterizable pre-filter. If the verification in step 66 has the result that the newly calculated parameterization differs from the current one by more than the predetermined threshold, the newly calculated filter coefficients and the newly calculated amplification value or the interference power limit are passed to buffer in node comparison means 26, for an interpolation to be discussed and the node comparison means 26 transfer to the pre-filter 30 the newly calculated filter coefficients in step 68 and the newly calculated amplification value in step 70. If, however, this is not the case and the newly calculated parameterization does not differ from the current one by more than the predetermined threshold, the node comparison means (26) will transfer to the pre-filter 30 in step 72, instead of the newly calculated parameterization, only the parameterization of current node, that parameterization that finally resulted in a positive value in step 66, ie it differs from a previous node parameterization in more than a predetermined threshold. After steps 70 and 72, the process of Figure 3 returns to process the next audio block, ie to a query 60. In the case that the newly calculated parameterization does not differ from the current node parameterization and consequently the pre -filter 30 in the lid 72 again obtains the node parameterization already obtained for at least the last audio block, the pre-filter 30 will apply this node parameterization to all the sample values of this audio block in the FIFO 32 , as will be described in greater detail below, which is how this current block is taken from the FIFO 32 and the quantizer 28 receives an audio block resulting from the pre-filtered audio values. Figure 4 illustrates the mode of operation of the parameterizable pre-filter 30 for the case that receives the newly calculated parameterization and the newly calculated amplification value, because they differ sufficiently from the current node parameterization in greater detail. As described with reference to Figure 3, there is no processing according to Figure 4 for each of the successive audio blocks, but only for the audio blocks where the restrictive parameterization differs sufficiently from the node parameterization current . The other audio blocks are as just described, pre-filtered when applying the respective current node parameterization and the respective current amplification value belonging to all the sample values of these audio blocks. In step 80, the parameterizable pre-filter 30 checks whether a transfer of newly calculated filter coefficients from the node comparison means 26 has been carried out or of older node parameterizations. The pre-filter 30 performs verification 80 until this transfer is carried out. As soon as the transfer has been carried out, the parameterizable pre-filter 30 starts to process the current audio value block of audio just in the buffer 32, that is to say the one for which the parameterization has just been calculated. In Figure 5a, for example, it is illustrated that all the audio values 56 against the audio value with the number 0 have already been processed and thus the memory 32 has already passed. The processing of the audio value block versus the Audio value with the number 0 is activated because the parameterization calculated for the audio block versus block 0, ie x0 (i), differs from node parameterization that is passed before the pre-filter 30 by more than the threshold predetermined. The parameterization x0 (i) in this way is a node parameterization as described in the present invention. The processing of the audio values in the audio block against the audio value 0 was made based on the set of parameters a0, x0 (i). It is considered in FIG. 5a that the parameterization was calculated for block 0 with the audio value 0 - 127 deferred by less than the predetermined threshold of the parameterization x0 (i) which refers to the front block. This block 0 is thus also removed from the FIFO 32 by the pre-filter 30, likewise processed with respect to all its sample values O-127 by the parameterization x0 (i) supplied in step 72, as indicated by the arrow 81 described by "direct application", and then passed to the quantizer 28. The parameterization calculated for the block 1 still located in the FIFO 32, however, differs in contrast according to the illustrative example of Fig. 5a, in more than the predetermined threshold of the parameterization x0 ( i) and thus in step 68 is passed to the pre-filter 30 as a parametrization ^ i), together with the amplification value a1 (step 70) and, applies, the interference power limit pertaining, where the indices of ayx in Fig. 5 will be an index for the nodes, as used in the interpolation that will be discussed below, which is performed with respect to the sample values 128 - 255 in block 1, symbolized by an arrow 82 and performed by the steps following step 80 in FIG. 4. The processing in step 80 in this manner will begin with the occurrence of the audio block with the number 1. At the time when the set of parameters is passed to t xx , only audio values 128-255, ie the current audio block after the last processed audio block 0 or the pre-filter 30, are in the memory 32. After determining the transfer of the node parameters xxd) in step 80, the pre-filter 30 determines the interference power limit q corresponding to the amplification value a1 in step 84. This can be carried out by the node comparison means 26 which pass this value to the pre-filter 30 or by the pre-filter 30. filter 30 again calculates this value as described above with reference to step 64. After that, an index j is initialized to a sample value in step 86 to signal to the oldest sample value remaining in the FIFO memory 32 or the first sample value of the current audio block "block 1", that is, in the present example of FIG. 5 the sample value 128. In step 88, the parameterizable pre-filter performs an interpolation between the filter coefficients x0 and x ^ where here, the parameterization x0 acts as a node in the node having the audio value number 127 of the previous block 0 and the parameterization xx acts as a node in the node that has 255 the audio value number of the current block 1. These audio value positions 127 and 255 will be subsequently referred to as node 0 and node 1, where the node parameterizations referring to the nodes in Fig. 5a are indicated by arrows 90 and 92. In the step 88, the parameterizable pre-filter 30 performs the interpolation of the filter coefficients x0, a between the two nodes in the form of a linear interpolation to obtain the filter coefficients interpolated in the sample position j, ie x (tj) (i ), i = 1 ... N. After that, that is, in step 90, the parameterizable pre-filter 30 performs an interpolation between the interference power limit q1 and q0 to obtain an interpolated interference power limit in the position of mu estra j, that is to say q (t3). In step 92, the parameterizable pre-filter 30 subsequently calculates the amplification value for the sample portion j based on the interpolated interference power limit and the quantization interference power and preferably also the interpolated filter coefficients, i.e. , for example depending on the quantization interference power root / qdj), where for this, reference is made to the explanations of step 64 of FIG. 3. In step 94, the parameterizable pre-filter 30 applies then the calculated amplification value and the filter coefficients interpolated to the sample value at the sample position j, to obtain a filtered sample value for this sample position, ie S '(tj). In step 96, the parameterizable pre-filter 30 then checks whether the sample position j has reached the current node, ie the node 1, in the case of Fig. 5a the sample position 255, ie the value of sample for which the parameterization transferred to the parameterizable pre-filter 30 plus the amplification value will be directly valid, that is, without interpolation. If this is not the case, the parameterizable pre-filter 30 will increase or increase the index j in 1, where steps 88-96 will be repeated. If the verification in step 96, however, is positive, the parameterizable pre-filter will apply, in step 100, the last amplification value transmitted from the comparison means of the node 26 and the last filter coefficients transmitted from the transmission means. comparison of node 26, directly without an interpolation to the sample value in the new node, whereby the current block, ie in the present case block 1, has been processed and the process is performed again in step 80 to the subsequent block to process that, depending on whether the parameterization of the next audio block 2 differs sufficiently from the parameterization x1 (i), this next audio block may be block 2 or otherwise a subsequent audio block. Before the additional procedure when processing the filtered sample values s' will be described with reference to Fig. 5, the purpose and background of the procedures of Figs. 3 and 4 are described below. The purpose of filtering is to filter the audio signal in the feed 12 with an adaptive filter, the transfer function of which is continuously adjusted to the inverse of the hearing threshold to the best possible degree, which also changes over time. The reason for this is that, on the decoder side, the reverse filtering of the transfer function of which is set to the hearing threshold forms the white quantization interference introduced by the quantization of the filtered audio signal, i.e. constant frequency quantization interference, by an adaptive filter, ie adjusts the same to the shape of the hearing threshold. The application of the amplification value in steps 94 and 100 in the pre-filter 30 is a multiplication of the audio signal or the filtered audio signal, i.e. the sample values or the filtered sample values s', by the amplification factor. The purpose here is to adjust the interference or quantization noise introduced into the audio signal filtered by the quantization described in greater detail below, and which is adjusted by the reverse filtering on the decoder side to the shape of the hearing threshold, as high as possible without exceeding the hearing threshold. This can be exemplified by the Parsevals formula according to which the square of the magnitude of a function is equal to the square of the magnitude of the Fourier transform. When on the decoder side the multiplication of the audio signal in the pre-filter by the amplification value is reversed again by dividing the filtered audio signal by the amplification value, the quantization interference power is also reduced, that is, by a factor a "2, a is the amplification value." Consequently, the quantization interference power can be adjusted to a high optimum degree by applying the amplification value in the pre-filter 30, which is synonymous with the size of the quantization stage increased and in this way the number of quantization steps to be encoded is reduced, which in turn increases the compression in the subsequent redundancy reduction part.Since differently, the effect of the pre-filter can be considered as a normalization of the signal to its masking threshold, such that the level of quantization interference or quantization noise it can stay constant both in time and frequency. Since the audio signal is in the time domain, quantization in this way can be performed step by step with uniform constant quantization, as will be described subsequently. In this way, ideally any possible irrelevance is eliminated from the audio signal and a disruptive or lossy compression scheme, it can be used to also remove the remaining redundancy in the pre-filtered and quantized audio signal, as will be described below. With reference to Fig. 5a, again it will be pointed out explicitly that of course the filter coefficients and the amplification values a0, alf x0, x employees must be available on the decoder side as lateral information, that the complexity of Transfer of this however is decreased by simply not using new filter coefficients and new amplification values for each block. In contrast, a threshold value check 66 is carried out in order to only transfer the parameterisations as lateral information with a sufficient parameterization change and otherwise not to transfer the lateral information or parameterisations.

An interpolation of the old parameterization to the new one is carried out in the audio blocks for which the settings have been transferred. The interpolation of the filter coefficients is carried out in the manner described above with reference to step 88. The interpolation with respect to the amplification is carried out by a deviation, that is by a linear interpolation 90 of the power limit of interference q0, qx. Compared to direct interpolation by the amplification value, linear interpolation results in better hearing or fewer audible artifacts with respect to the interference power limit. Subsequently, the further processing of the pre-filtered signal will be described with reference to Fig. 6, which basically includes quantification and reduction of redundancy. First, the filtered sample values that are sent out by the parameterizable pre-filter 30, are stored in the buffer 38 and at the same time allow passage from the buffer 38 to the multiplier 40 where they are, since it is their first step, at the beginning they are passed without change, ie with a scale factor of 1, by the multiplier 40 to the quantizer 28. There, the audio values filtered on an upper limit are cut in step 110 and then quantized in the step 112. The two steps 110 and 112 are executed by the quantizer 28. In particular, the two steps 110 and 112 are preferably executed by the quantizer 28 in one step by quantization of the filtered audio values s' by a function of quantization step which maps the filtered sample values s' exemplary present in a floating point illustration to a plurality of integer quantization stage indices or values and which has a quant The flat rate for the sample values is filtered from a certain threshold value onwards, such that the filtered sample values greater than the threshold value are quantized at one and the same quantization stage. An example of this quantization stage function is illustrated in Fig. 7a. The quantized filtered sample values are referred by s' in Fig. 7a. The function of the quantization stage of preference is a quantization stage function with a stage size that is constant below the threshold value, ie the jump to the next quantization stage will always be carried out after a constant interval on the power values S '. In the implementation, the stage size at the threshold value is adjusted, such that the number of quantization stages of preference corresponds to a power of 2. Compared to the floating-point illustration of the filtered sample values of input s', the threshold value is smaller such that a maximum value of the illustrable region of the floating point illustration exceeds the threshold value. The reason for this threshold value is that it has been observed that the output of the audio signal filtered by the pre-filter 30, occasionally includes audio values that contribute very large values due to an unfavorable accumulation of harmonic waves. Furthermore, it has been found that cutting these values, as achieved by the function of the quantization stage shown in Fig. 7a, results in a high data reduction, but only in a minor deterioration of the audio quality. In contrast, these occasional locations in the filtered audio signal are artificially formed by frequency selective filtering in the parameterizable filter 30, such that its cutoff deteriorates the audio quality only in a minor proportion. A somewhat more specific example of the quantization stage function shown in Fig. 7a would be one that rounds off all filtered sample values s' to the next integer up to the threshold value, and hence quantifies all filtered sample values prior to the stage of highest quantification, such as for example 256. This case is illustrated in Fig. 7a. Another example of a possible quantization stage function would be that shown in Fig. 7b. Up to the threshold value, the function of the quantization stage of Fig. 7b corresponds to Fig. 7a. Instead of having an abruptly flat course for the sample values s' over the threshold value, however, the quantization stage function continues with a smaller slope than the slope in the region below the threshold value. Stated differently, the size of the quantization stage is larger than the threshold value. With this, a similar effect is achieved by the quantization function of Fig. 7a, but on the one hand, with greater complexity due to the different stage sizes of the quantization stage function on and below the threshold value and by another part, improved audio quality, since very high filtered audio values s' do not cut off completely but only quantify with a higher quantization stage. As already described above, on the decoder side not only the quantized and filtered audio values s' must be available but also the feed parameters for the prefilter 30 are the filtering base of these values, it is say the node parameterization including an indication to the belonging amplification value. In step 114, the compressor 34 thus performs a first compression test and thus compresses lateral information containing the amplification values a0 and a1 in the nodes such as for example 127 and 255, and the filter coefficients x0 and xx in the nodes and by the filtered sample values quantized s' to a temporal filtered signal. The compressor 34 in this manner is a lossy operation encoder such as for example a Huffman or arithmetic encoder with or without prediction and / or adaptation. The memory 38 which passes the sampled audio values s 'serves as a buffer for a convenient block size with which the compressor 34 processes the quantized, filtered and also scaled audio values as previously described s' output by the quantizer 28. The block size may differ from the block size of the audio blocks as used by the means 20. As already mentioned, the bit rate controller 36 has the multiplexer 40 controlled by a multiplying of 1 for the first compression test, such that the filtered audio values pass unchanged from the pre-filter 30 to the quantizer 28 and hence as quantized filtered audio values to the compressor 34. The compressor 34 monitors in step 116 if a certain compression block size, ie a certain number of quantized sampled audio values, has been coded into the temporal encoded signal, or if The filtered quantized audio values s' will be encoded in the current temporal encoded signal. If the compression block size has not been reached, the compressor 34 will continue to perform the current compression 114. If the compression block size has however been reached, the bit rate controller 36 will verify in step 118 whether the amount of bits required for compression is greater than a number of bits dictated by a desired bit rate. If this is not the case, the bit rate controller 36 will verify in step 120 whether the number of bits required is smaller than the number of bits dictated by the desired bit rate. If this is the case, the bit rate controller 36 will fill the signal encoded in step 122 with padding bits until the number of bits dictated by the desired bit rate has been reached. Subsequently, the encoded signal is outputted in step 124. As an alternative to step 122, the bit rate controller 36 can pass the compression block of the filtered audio values s' still stored in the memory 38 wherein the last compression is based on a form multiplied by a multiplier greater than 1 by the multiplier 40 to the quantizer 28 to again pass the steps 110-118 until the number of bits dictated by the desired bit rate has been reached, as indicated by a step 125 illustrated with dotted lines. However, if the check in step 118 results in the amount required being greater than that dictated by the desired bit rate, the bit rate controller 36 will change the multiplier by the multiplier 40 to a factor between 0 and 1 exclusive. This is done in step 126. After step 126, the bit rate controller 36 provides the memory 38 for outputting again the last compression block of the filtered audio values s' in which it is stored. has based the compression, where subsequently multiplied by the factor set in step 126 and again supplied to the quantizer 28, whereby steps 110-118 are performed again and until then the temporarily encoded signal is discarded. It should be noted that when steps 110-116 are performed again, in step 114 of course the factor employed in step 126 (or step 125) is also integrated into the encoded signal. The purpose of the method after step 126 is to increase the effective stage size of the quantizer 28 by the factor. This means that the resulting quantization interference is uniformly over the masking threshold, which results in audible interference or audible noise, but results in a reduced bit rate. If, after passing steps 110-116 again, it is determined again in step 118 that the number of bits required is greater than dictated by the desired bit rate, the factor will be reduced again in step 126, etc. If the data is finally outputted in step 124 as an encoded signal, the next compression block will be made from the subsequent quantized filtered audio values s'. It should also be noted that another pre-initialized value different from 1 may be used as the multiplication factor, ie for example 1.

Scaled adjustment will then be carried out in any case at the beginning, ie in the upper part of Fig. 6. Fig. 5b illustrates again the resulting coded signal which is generally indicated by 130. The encoded signal includes information lateral and main intermediate data. The lateral information includes, as already mentioned, information of which for special audio blocks, ie audio blocks where a significant change in the filter coefficients has resulted in the audio block sequence, the amplification value and the value of the filter coefficients has resulted in the sequence of audio blocks, the amplification value and the value of the filter coefficients can be derived. If necessary, the lateral information will include more information regarding the amplification value used for the bit controller. Due to the mutual dependence of the amplification value and the interference power limit q, the lateral information may optionally, apart from the amplification value a # to a node #, include the interference power limit q #, or only the latter. The lateral information preferably is arranged within the encoded signal, such that information lateral to the filter coefficients and the corresponding application value, or corresponding interference power limit is set against the main data to the audio block of quantized filtered audio values s', of which these filter coefficients with corresponding amplification values or corresponding interference power limit have been derived, ie lateral information a0, x0d) after block -1 and lateral information ax , xx (i) after block 1. Said in a different way, the main data, that is, the quantized filtered audio values s', starting from, excluding, an audio block of the type where a significant change in the sequence of audio blocks has resulted in the filter coefficients, up to, including, the next audio block of this type in Fig. 5, for example lo, the audio values s '(t0) s' (t2SS), will always be arranged between the lateral information block 132 to the first of these two audio blocks (block -1) and the other lateral information block 134 to the second of the two audio blocks (block 1). The audio values s '(t0) -s' (t127) which are decodable or have been, as previously mentioned with reference to Fig. 5a, obtained only by lateral information 132, while the audio values s '(t128) -' (t25S) have been obtained by interpolation by lateral information 132 as support values at the node with the sample value number 127 and by lateral information 134 as support values at the node with the sample value number 255 and in this way they are decoded only by both lateral information. In addition, the lateral information regarding the amplification value or the interference power limit and the filter coefficients in each lateral information block 132 and 134, are not always integrated independently of each other. On the contrary, this lateral information is transferred in differences to the previous lateral information block. In FIG. 5b for example, the lateral information block 132 contains the amplification value a0 and the filter coefficients x0 with respect to the node at time t.x. In the lateral information block 132, these values can be derived from the block itself. From the side information block 134 however, the lateral information regarding the node at time t2SS may no longer be derived from this block alone. In contrast, the lateral information block 134 only includes difference information of the amplification value ax of the node at time t255 and the amplification value of the node at time t0 and the differences of the filter coefficients xx and the filter coefficients x0. The lateral information block 134 consequently only contains the information in ax - a0 and xx (i) - x0 (i). At intermittent times however, the filter coefficients and the amplification value or the interference power limit should be transferred completely and not only as a difference to the previous node, such as for example every second to allow a receiver or decoder to engage in a flow of decoding data, as will be discussed below. This type of lateral information integration in lateral information blocks 132 and 134 offers the advantage of the possibility of a higher compression speed. The reason for this is that, although lateral information if possible, will only be transferred if a sufficient change of the filter coefficients to the filter coefficients of a previous node has resulted, the complexity of calculating the difference on the encoder side or calculating the sum on the decoder side rewards since the resulting differences are small despite the query in step 66 to thereby allow advantages in entropy coding. After a modality of an audio encoder has been previously described, a mode of an audio decoder that is suitable for decoding the encoded signals generated by the audio encoder 10 of FIG. 1 to a decodable or reproducible audio signal decoded , will be described subsequently. The configuration of this decoder is illustrated in Fig. 8. The decoder generally indicated by 210 includes a decompressor 212, a FIFO memory 214, a multiplier 216 and a parameterizable post-filter 218. The decompressor 212, the FIFO memory 214, the multiplier 216 and the parameterizable post-filter 218 are connected in this order between a data feed 220 and a data output 222 of the decoder 210, wherein the encoded signal is received in the data feed 220 and the decoded audio signal only differs from the original audio signal in the data feed 12 of the audio encoder 10 by the quantization noise generated by the quantizer 28 in the audio encoder 10, is output from the data output 222. The decompressor 212 is connects to a control supply of the multiplier 216 in another data output to pass a multiplying to it, and to a parametrization feed of the post-filter param etridable 218 by another data output. As illustrated in FIG. 9, decompressor 212 initially decompresses the signal compressed in data feed 220 in step 224 to obtain quantized filtered audio data, i.e. sample values s', and lateral information corresponding to the side information blocks 132, 134 which, as illustrated, indicate the amplification and filter coefficients or instead of the amplification values, the interference power limits at the nodes. As shown in Fig. 10, decompressor 212 checks the decompressed signal in the order of appearance in step 226 if the lateral information with filter coefficients is contained, in a self-contained form without a difference reference to a block of previous lateral information. Stated differently, the decompressor 212 searches for the first lateral information block 132. As soon as the decompressor 212 has found something, the quantized filtered audio values s' pass from the buffer to the FIFO memory 214 in step 228. If a complete audio block of quantized filtered audio values s' has been stored during step 228 without a directly following side information block, it will initially be post-filtered in step 228 by the information contained in the side information received in step 226, in the parameterization and amplification value in a post-filter and amplify in the multiplier 216, which is how it is decoded and in this way the corresponding decoded audio block is achieved. In step 230, the decompressor 212 checks the decompressed signal by the occurrence of any type of lateral information block, that is, with absolute filter coefficients or differences in filter coefficients with a previous lateral information block. In the example of Fig. 5B, decompressor 212 for example will recognize the occurrence of side information block 134 in step 230 before recognition of side information block 132 in step 226. In this way, the block of audio values quantized filters s '(t0) -s' (t127) will be decoded in step 228, using lateral information 132. As long as the side information block 134 in the decompressed signal has not yet occurred, the buffering and probably decoding of the blocks is continued in step 228 by the lateral information of step 226, as previously described. As soon as the lateral information block 132 has occurred, the decompressor 212 will calculate the parameter values at node 1, i.e. ax, xa (i), at step 232 when adding the difference values in the lateral information block 134 and the parameter values in the side information block 132. Step 232 is of course omitted if the current lateral information block is a self-contained side information block with no differences, which as previously described may occur in exemplary manner. every second. In order that the waiting time for the decoder 210 is not very long, the side information blocks 132 in which the parameter values can be absolutely derived, ie without relation to another block of lateral information, are arranged at sufficiently small distances in such a way that the activation time or the non-operating time when the audio encoder 210 is switched in the chaos of, for example, a broadcast transmission or radio transmission, is not very large. Preferably, the number of lateral information blocks 132 arranged intermediate with the difference values are arranged at a fixed predetermined number between the lateral information blocks 132, such that the decoder knows when a type of lateral information block 132 of new will be expected in the encoded signal. Alternately, different types of lateral information blocks are indicated by corresponding flags.

As illustrated in Fig. 11, after a side information block for a new node has been reached, in particular after step 226 or 232, a sample value index j at the beginning is initialized to 0 in the step 234. This value corresponds to a sample position of the first sample value in the audio block currently remaining in the FIFO 214 to which the current lateral information refers. Step 234 is performed by the parameterizable post-filter 218. Post-filter 218 then calculates the interference power limit at the new node in step 236, where this step corresponds to step 84 of FIG. 4 and it can be omitted when, for example, the interference power limit at the nodes is transmitted in addition to the amplification values. In subsequent steps 238 and 240, the post-filter 218 performs interpolations with respect to the filter coefficients and the interference power limit corresponding to the interpolations 88 and 90 of FIG. 4. The subsequent calculation of the amplification value for the sample position j based on the interpolated interference power limit and the interpolated filter coefficients of steps 238 and 240 in step 242, corresponds to step 92 of FIG. 4. In step 244, the post filter 218 applies the value of amplification calculated in step 242 and the filter coefficients interpolated to the sample value in the sample position j. This step differs from step 94 in FIG. 4 by the fact that the interpolated filter coefficients are applied to the filtered sample values quantized s' such that the transfer function of the parameterizable post-filter does not correspond to the inverse of the hearing threshold, but at the hearing threshold itself. In addition, the postfilter does not multiply by the amplification value but a division by the amplification value in the quantized filtered sample values s' or the quantized filtered sample value, already inverse filtering in the j position. If the post-filter 218 has not yet reached the current node with the sample position j, which it verifies in step 246, it will increase the index of the sample position j in step 248 and start steps 238-246 again. Only when the node has been reached will it apply the amplification value and the filter coefficients of the new node to the sample value in the node, ie in step 250. The application in turn includes, as in step 218, a division by the value of amplification and filtering with a transfer function that equals the hearing threshold and not the inverse of the latter instead of a multiplication. After step 250, the current audio block is decoded by an interpolation between two node settings. As already mentioned, the interference introduced by the quantization when encoded in step 110 or 112, is adjusted both in form and magnitude with the hearing threshold by the filtering and the application of an amplification value in steps 218 and 224. It should also be noted that in the case that the quantized filtered audio values have been subjected to another multiplication in step 126 due to the bit rate controller before being coded in the coded signal, that factor can also be considered in the steps 218 and 224. Alternatively, the audio values obtained by the process of FIG. 11 can of course be subjected to another multiplication to correspondingly amplify again the weakened audio values by a lower bit rate. With respect to Figs. 3, 4, 6, and 9-11, it is noted that they show flow charts illustrating the mode of operation of the encoder of Fig. 1 or the decoder of Fig. 8, and that each of the illustrated stages in the flowchart for a block as described, it is implemented in corresponding media as previously described. The implementation of the individual stages can be achieved in physical equipment, as a part of the ASIC circuit or in software, as a subroutine. In particular, the explanations described in the bl in these Figures approximately indicate to which process the respective stage corresponding to the respective blrefers, while the arrows between the bl illustrate the order of the steps when the encoder and decoder are operated, respectively. With reference to the previous description, it is again pointed out that the coding scheme illustrated above may vary in many aspects. In an exemplary manner, it is not necessary that an amplification and parameterization value or an interference power limit, as determined for a certain audio bl be considered as directly valid for a certain audio value, as in the previous mode the last respective audio value of each audio bl ie the value 128 ° in this audio blin such a way that the interpolation for this audio value can be omitted. On the contrary, it is possible to relate these values of node parameters to a node that is temporarily between sample times tn, n = 0, ..., 127, of the audio values of this audio bl in such a way that an interpolation will be necessary for each audio value. In particular, the parameterization determined by an audio blor the amplification value determined by this audio blcan also be indirectly applied to another value, such as, for example, the audio value in the middle of the audio bl such as for example the 64th audio value in the case of the previous blsize of 128 audio values. Additionally, it is pointed out that the above embodiment refers to an audio coding scheme designed to generate a coded signal with a controlled bit rate. Controlling the bit rate however is not necessary for every application case. This is the reason why the corresponding steps 116 to 122 and 126 or 125 can also be omitted. With reference to the compression scheme mentioned with reference to step 114, for reasons of integrity, reference is made to the document by Schuller et al. , described in the introduction to the description and in particular to the division IV, the contents of which with respect to reducing redundancy by means of loss coding, are incorporated herein by reference.

The following is indicated with reference to the above description. Although the present invention has been described above with reference to a special audio coding scheme that allows for shorter delay times, the present invention can of course also be applied to different audio encodings. In exemplary form, an audio coding scheme will be conceived where the encoded signal consists of the highly quantized audio values filtered without a reduction of redundancy to be made. Correspondingly, also it is designed to perform selective frequency filtering differently from the manner described above, ie on the encoder side with a transfer function that equals the inverse of the threshold of hearing and the decoder side with a transfer function that matches the hearing threshold. Additionally, individual aspects of the above modalities may be omitted. In this way, for example, it is possible when the compression ratio to transmit the lateral information referring to each audio block is reduced, omitting interpolation and / or always transferring the parameters in the lateral information in blocks of self-contained information and not as differences. referring to the previous lateral information blocks. Additionally, the present invention is not limited to audio signals. It can also be applied to different information signals, such as for example video signals consisting of a sequence of frames, ie a sequence of pixel sets. In any case, the above audio coding scheme provides a way to limit the bit rate in an audio encoder with very short delay time. The bit rate generates peaks that result when coding of the audio signal is dependent, which are avoided by limiting the initial value range of the pre-filter. Since consequently corresponds to the nature of audio signals transferred resulting in speeds differently high bit transfer, ie signals more complex audio resulting in speeds higher bits and less complex resulting in lower bit , an upper limit for the bit rate of the transfer, which for example often exists in wireless transfer means, can always be satisfied. The change in the function of the quantization stage over the threshold is a convenient means to limit the bit rate to the maximum allowed. In the above embodiments, the encoder has included a pre-filter which forms the audio signal Conveniently, a quantizer having a cap size quantization, followed by an entropy encoder, the quantizer has generated values also refer as indexes. In general, high rates also signify a higher bit rate connected thereto, which, however, has been avoided by limiting (Fig. 7a) or thin (Fig. 7b) the index range, however it involves the possibility of deterioration the audio quality. In addition, the following is indicated with reference to the previous modality. Although previously described that the threshold value remains constant when quantifies or even function of the quantization step remains constant, ie artifacts generated in the filtered audio signals are always quantized or cut by a coarser quantization, which can deteriorate audio quality to an audible extent, it is also possible to only use these measures if the complexity of the audio signal requires this, ie if the bit rate required for coding exceeds a desired bit rate.

In this case, in addition to the functions of the quantization stage shown in Figures 7a and 7b, for example with a constant quantization stage size over the entire range of possible values at the pre-filter output, could be used and the quantifier for example will respond to a signal to use either the function of the quantization stage with an always constant quantization stage size or one of the quantization stage functions according to Figures 7a or 7b of such that the quantifier can be told by the signal that it makes, with little deterioration of the audio quality, the decrease in the quantization stage over the threshold value or cut-off over the threshold value. Alternatively, the threshold value can also be gradually reduced. In this case, the reduction in threshold value can be performed in place of the factor reduction of step 126. After a first compression test without step 110, the temporarily compressed signal can only be subjected to a selective threshold value quantification in a modified step 126 if the bit rate was still too high (118). In another step, the filtered audio values will then be quantized with the function of the quantization stage having a flatter course over the audio threshold. In addition, reductions in bit rate can be made in the modified step 126 by reducing the threshold value and in this way by another modification of the function of the quantization stage. In particular, it is pointed out that, depending on the circumstances, the audio coding scheme of the invention can also be implemented in software. The implementation can be in a digital storage medium, in particular on a disk or on a CD having control signals, which can be read electronically, which can cooperate with a programmable computer system in such a way that the corresponding method is will execute In general, the invention also resides in a computer program product, which has a program code stored in a machine readable carrier to perform the method of the invention when the computer program product runs on a computer. Stated differently, the invention can also be achieved as a computer program having a program code to perform the method when the computer program runs on a computer. In particular, the above method steps in the flowchart blocks can be implemented individually or in groups of several together in subprogram routines. Alternatively, an implementation of a device of the invention in the form of an integrated circuit, of course, is also possible where these blocks, for example, are implemented as part of the individual circuit of an ASIC. In particular, it is pointed out that depending on the circumstances, the scheme of the invention can also be implemented in software or program. The implementation can be in a digital storage medium, in particular on a disk or CD having control signals that can be read electronically, which can cooperate with a programmable computer system in such a way that the corresponding method will be executed. In general, the invention in this way also resides in a computer program product having a program code stored in a machine readable carrier to perform the method of the invention when the computer program runs on the computer. Stated differently, the invention can also be achieved as a computer program that has a program code to perform the method when the computer program runs on a computer.

Claims (10)

1. CLAIMS 1. A device for quantizing an information signal of a sequence of information values, the information signal is an audio signal and the information values are audio values, characterized in that it comprises: means for determining a hearing threshold for a block of audio values and a sequence of audio values; means for calculating a version of a parameterisation of a parameterizable filter, in such a way that its transfer function corresponds approximately to the inverse of the magnitude of the first hearing threshold; means for frequency selective filtering of the sequence of audio values and for obtaining a sequence of filtered audio values; means for quantifying the filtered audio values, to obtain a sequence of quantized audio values by means of a quantization stage function that maps the filtered audio values to the quantized audio values and the course of which is steepest below a information threshold value that on threshold information value; wherein the means for frequency selective filtering comprise: means for filtering a predetermined block of audio values of the audio value sequence with the parameterizable filter using a predetermined parameterization depending on a predetermined form of the parameterization version to obtain a block of filtered audio values. The device according to claim 1, characterized in that the means for determining a hearing threshold are formed to additionally determine another second hearing threshold for another second block of audio values, and the means for calculation are formed to calculate a version of another second parametrisation of the parameterizable filter, in such a way that its transfer function corresponds approximately to the inverse of the magnitude of the second hearing threshold, wherein the means for selective frequency filtering comprise: means for interpolating between the version of the first parameterization and the version of the second parameterization, to obtain a version of an interpolated parameterization for a predetermined audio value of the predetermined block of audio values; and means for applying the version of the interpolated parameterization to the predetermined audio value of the block of predetermined audio values. The device according to claim 2, further comprising means for determining a first interference power limit that depends on the first masking threshold and a second interference power limit that depends on the second masking threshold, and wherein the filtering means comprises means for interpolating between the first interference power limit and the second interference power limit, to obtain an interpolated interference power limit for a predetermined audio value of the predetermined block of audio values, means to determine an intermediate scaling value that depends on a quantizing interference power caused by the quantization according to a predetermined quantization rule and the interpolated interference power limit and means for applying the adjustment value on an intermediate scale to the default audio value or to obtain a filtered audio value adjusted in scale. The device according to claim 3, characterized in that the means for interpolating between the first interference power limit and the second interference power limit perform a linear interpolation. The device according to claim 3 or 4, characterized in that the means for determining the adjustment value in the intermediate scale comprises means for calculating the root of the quotient of the quantization interference divided by the interpolated interference power limit. The device according to one of the preceding claims, characterized in that the means for quantizing are formed to perform quantization in response to a control signal. The device according to one of the preceding claims, characterized in that it also comprises lossless compression means for compressing the filtered audio values into a compressed audio stream, wherein the compression means is formed to control a bit rate. of the compressed audio stream and for sending the control signal to the means for quantizing in the event that the bit rate is greater than a control value. The device according to one of the preceding claims, characterized in that the function of the quantization stage has a flat course on the threshold information value, such that the filtered audio values greater than the threshold information value are quantify to a value of the maximum quantification stage. 9. A method for quantizing an information signal of a sequence of information values, the information signal is an audio signal and the information values are audio values, comprising the steps of: frequency selective filtering for the sequence of values audio, to obtain a sequence of filtered audio values; quantify the filtered audio values to obtain a sequence of quantized audio values by means of a quantization stage function that maps the filtered audio values to the quantized audio values and the course of which is steepest below an information value threshold than on the threshold information value; determine a hearing threshold for a block of audio values; and calculating a version of a parametrisation of a parameterizable filter, in such a way that its transfer function corresponds approximately to the inverse of the magnitude of the first hearing threshold, wherein the stage of selective frequency filtering also includes the step of: filtering a predetermined block of audio values of the sequence of audio values with the parameterizable filter using a predetermined parameterization which in a predetermined form depends on the version of the parameterization to obtain a block of filtered audio values. 10. A computer program having a program code for performing the method according to claim 9, when the computer program is run on a computer.