Encoding system comprising a subband coder for subband coding of a wideband digital signal constituted by first and second signal compon

We claim:

1. An encoding system, for encoding a digital information signal having a first signal component and a second signal component, comprising

subband coding means for generating (a) a first subband signal in response to the first signal component, the first subband signal including a first signal block of q samples of the first subband signal, where q is a positive integer, and (b) a second subband signal in response to the second signal component, the second subband signal including a second signal block of q samples of the second subband signal, the first and the second subband signals being in a same subband and the first and second signal blocks being time-equivalent; and

quantizing means for quantizing said samples in said first and second signal blocks to form first and second quantized signal blocks, said samples in said first and second quantized signal blocks being represented by n.sub.m1 and n.sub.m2 bits respectively, and

said quantizing means comprises:

bit need means for determining a respective bit need b for said first signal block and a respective bit need b for said second signal block, said respective bit need for the first signal block being related to the number of bits by which the samples in said first signal block should be represented, and said respective bit need for the second signal block being related to the number of bits by which the samples in said second signal block should be represented,

bit allocation means, responsive to the respective bit needs b for said first signal block and for said second signal block determined in the bit need means, for allocating a portion of an available quantity of bits to said samples in said first and second signal blocks to obtain the values n.sub.m1 and n.sub.m2 for the corresponding quantized first and second signal blocks,

determining means for determining, for one block of said first and second signal blocks, whether an initial allocation of bits to the samples in said one block is to be performed in the bit allocation means irrespective of the respective bit need b of said one block determined by said bit need means; and

means for producing a first control signal responsive to the determination by said determining means,

said bit allocation means being responsive to said first control signal, for initially allocating a first number of bits to the samples of said one block, and for initially allocating a second number of bits to the samples of the other block of said first and second signal blocks, irrespective of the respective bit need b of said other block determined by said bit need means.

2. A system as claimed in claim 1, characterized in that said first and said second numbers are the same.

3. A system as claimed in claim 1, characterized in that said determining means evaluates each of said signal blocks to determine if said initial allocation is to be performed for at least one of said signal blocks, and

said means for producing produces said first control signal if said initial allocation is to be performed for at least one of said signal blocks.

4. A system as claimed in claim 3, characterized in that said determining means comprises means for determining, for said other block, whether no bit allocation to the samples in said other block should be allocated in the bit allocation means, and

means for producing a second control signal in response to said determining means determining that, for each of said first and second signal blocks, no initial allocation is to be performed, and that no bit allocation to the samples in said other block should be allocated in the bit allocation means, and

said bit allocation means comprises means, responsive to said second control signal, for allocating no bits to said other block.

5. A system as claimed in claim 3, characterized in that said determining means comprises first means for determining, for said one block, whether no bit allocation to the samples in said one block need be allocated in the bit allocation means; and second means for determining, for said other block, whether no bit allocation to the samples in said other block need be allocated in the bit allocation means, and

means, responsive solely to said first and second means each determining, for the respective one and other blocks, that no bit allocation need be performed, for producing a second control signal, and

said bit allocation means comprises means, responsive to said second control signal, for allocating no bits to either of said first and second signal blocks.

6. An encoding system comprising a subband coder for subband digital coding of an information signal constituted by at least first and second signal components, said subband coder including:

signal splitting means, responsive to said information signal, for splitting said information signal into M successive subbands and generating digital subband signals comprising a succession of signal blocks, in a subband SB.sub.m a succession of first and second corresponding time-equivalent signal blocks respectively containing samples of first and second subband signal components related to said first and second signal components, and

quantizing means for quantizing block-by-block said samples of first and second subband signal components in the subband SB.sub.m to form quantized signal blocks, said samples in one pair of corresponding quantized signal blocks in said subband SB.sub.m being represented by n.sub.m1 and n.sub.m2 bits respectively,

said quantizing means comprising:

bit need means for determining a respective bit need b for respective signal blocks in the M subbands, said respective bit need being related to the number of bits by which said samples in said signal blocks in respective ones of the M subbands should be represented,

bit allocation means, responsive to the bit needs determined in the bit need means, for allocating an available quantity of bits to respective samples in the different signal blocks in the M subbands,

determining means for determining, for one signal block of said pair, whether an initial allocation of bits to the samples in said one signal block is to be performed in the bit allocation means irrespective of the bit need of said one signal block determined by said bit need means; and

means for producing a first SB.sub.m control signal responsive to the determination by said determining means,

said bit allocation means being responsive to said first SB.sub.m control signal, for initially allocating a first number of bits to the samples of said one signal block, and for initially allocating a second number of bits to the samples of the other block of said pair irrespective of the bit need of said other block determined by said bit need means.

7. A system as claimed in claim 6, wherein said first number and said second number are the same.

8. A system as claimed in claim 7, characterized in that said determining means evaluates each signal block of said pair to determine if said initial allocation is to be performed for one of said signal blocks of said pair,

said means for producing produces said first SB.sub.m control signal if said initial allocation is to be performed for said one signal block of said pair,

said determining means further comprises means for determining, for the other signal block of said pair, that no bit allocation to the samples in said other signal block should be allocated in the bit allocation means, and

means for producing a second SB.sub.m control signal in response to said determining means determining that, for each signal block of said pair, no initial allocation is to be performed, and that, for said other signal block, no bit allocation to the samples in that signal block should be allocated in the bit allocation means, and

said bit allocation means comprises means, responsive to said second SB.sub.m control signal, for allocating no bits to said other signal block irrespective of the bit need of said other signal block.

9. A system as claimed in claim 7, characterized in that said determining means evaluates each signal block of said pair to determine if said initial allocation is to be performed for at least one of said signal blocks of said pair, and

said means for producing produces said first SB.sub.m control signal if said initial allocation is to be performed for at least one signal block of said pair.

10. A system as claimed in claim 9, characterized in that said determining means comprises first means for determining, for said one signal block of said pair, that no bit allocation to the samples in said one signal block need be allocated in the bit allocation means; and second means for determining, for the other signal block of said pair, that no bit allocation to the samples in said other signal block need be allocated in the bit allocation means,

means for producing a second SB.sub.m control signal, responsive solely to said first and second means each determining, for the respective block of said pair, that no initial allocation is to be performed, and

said bit allocation means comprises means, responsive to said second SB.sub.m, control signal, for allocating no bits to either signal block of said pair irrespective of the bit need of either signal block

11. An encoding system comprising a subband coder for subband digital coding, including coding in an intensity mode, of an information signal constituted by at least first and second signal components, including:

signal splitting means, responsive to said information signal, for splitting said information signal into M successive subbands and generating digital subband signals comprising a succession of signal blocks, in a subband SB.sub.m a succession of first and second corresponding time-equivalent signal blocks respectively containing samples of first and second subband signal components related to said first and second signal components,

means for combining corresponding samples of the first and second subband signal components of a subband SB.sub.p, where p.noteq.m, to obtain a combined subband signal, and quantizing the combined subband signal to provide a quantized combined subband signal constituted by combined signal blocks each containing q samples, the q samples in a combined signal block of the quantized combined subband signal each being represented by n.sub.pc bits, and

quantizing means for quantizing block-by-block the samples of first and second subband signal components in the subband SB.sub.m to form component signal blocks, the samples in a pair of corresponding component signal blocks of the first and second quantized subband signal components in said subband SB.sub.m being represented by n.sub.m1 and n.sub.m2 bits respectively,

wherein said quantizing means comprises:

bit need means for determining a respective bit need b for respective signal blocks in the subbands, said bit need being related to the number of bits by which the samples in a signal block in a respective subband should be represented,

bit allocation means, responsive to the bit needs determined in the bit need determining means, for allocating an available quantity of bits to respective samples in the different signal blocks in the M subbands, and

determining means for determining, for a particular signal block of the first subband signal component in the subband SB.sub.m, whether an initial allocation of a number of bits to the samples in said particular signal block is to be performed in the bit allocation means irrespective of the bit need b of particular signal block determined by said bit need means,

and wherein:

for the subband SB.sub.p, said determining means evaluates each signal block of an SB.sub.p pair of corresponding time-equivalent signal blocks prior to combining of the corresponding samples, and said means for producing produces a first SB.sub.p control signal if said initial allocation is to be performed for one signal block of said SB.sub.p pair,

said means for combining and quantizing then combines corresponding samples of the first and second subband signal components in the subband SB.sub.p to obtain said combined subband signal which is then quantized, and

said bit allocation means, responsive to said first SB.sub.p control signal, initially allocates a third number of bits to the samples of said combined subband signal.

12. A system as claimed in claim 11, characterized in that for subband SB.sub.p said determining means comprises first means for determining, for said one signal block of said SB.sub.p pair, that no bit allocation to the samples in said one signal block need be allocated in the bit allocation means; and second means for determining, for the other signal block of said SB.sub.p pair, that no bit allocation to the samples in said other signal block need be allocated in the bit allocation means,

means for producing a second SB.sub.p control signal, responsive solely to said first and second means each determining, for the respective signal block of said SB.sub.p pair, that no initial allocation is to be performed, and

said bit allocation means comprises means, responsive to said second SB.sub.p control signal, for allocating no bits for quantizing the combined subband signal for said SB.sub.p pair.

Description Submit all comments and votes

BACKGROUND OF THE INVENTION

The invention relates to an encoding system comprising a subband coder for subband coding of a wideband digital signal, for example, a digital stereo audio signal, constituted by at least first and second signal components which are sampled each with a specific sampling frequency F.sub.3, the subband coder including signal splitting means for generating, in response to the wideband digital signal, a number of M subband signals by means of a sampling frequency reduction, for which purpose the splitting means split up the wideband signal into successive subbands having band numbers m which augment with frequency, where m=1.ltoreq.m.ltoreq.M, and in which each subband signal is constituted by at least first and second subband signal components, the encoding system further including quantizing means for quantizing block-by-block the first and second subband signal components in a subband SB.sub.m, a quantized subband signal component comprising successive signal blocks, each signal block containing q samples, the q samples in corresponding signal blocks of the first and second quantized subband signal components in the subband SB.sub.m being represented by n.sub.m1 and n.sub.m2 bits respectively, the quantizing means to this end comprising bit need determining means for determining a bit need b for corresponding signal blocks in the subbands, which bit need is related to the number of bits by which the samples in a signal block in a subband SB should be represented, and the quantizing means including bit allocation means for allocating the available quantity of bits to samples in the different signal blocks in the subbands in response to the bit needs as they are determined in the bit need determining means so as to obtain the values n.sub.m1 and n.sub.m2 for the corresponding signal blocks in the subband SB.sub.m.

Such an encoding system is known from European Patent Application No. 289.080 (to which U.S. Pat. No. 4,896,362 corresponds) document (1) in the reference list. Dutch Patent Application No. 90.01.127 (to which U.S. patent application Ser. No. 07/620,971 corresponds, abandoned; and U.S. patent application Ser. No. 08/144,092 which is an indirect continuation thereof, corresponds), document (6a) in the reference list, further describes how the bit allocation in such an encoding system may be realised. A single subband signal in each of the subbands, for example, a mono signal or two or more signals in each of the subbands may be concerned. With two signals one may think of a stereo signal, with three signals one may think of a left, central, right signal and with four signals one may think of a quadraphonic signal.

SUMMARY OF THE INVENTION

It is an object of the invention to improve the bit allocation for the coding of a subband signal constituted by two or more subband signal components in each of the subbands.

According to the invention the encoding system is thereto characterized in that the quantizing means comprise determining means, which determining means are arranged for determining for a signal block of the first subband signal component in the subband SB.sub.m , whether an initial allocation of a number of bits to the samples in the signal block is to be performed in the bit allocation means irrespective of the bit need belonging to the signal block and determined in the bit need determining means, and are arranged for producing a first control signal in response thereto, and in that the bit allocation means are arranged for initially allocating a number of bits to the samples of the signal block in response to the first control signal, and also for initially allocating a number of bits to the samples of the corresponding signal block of at least the second subband signal component in the subband SB.sub.m, irrespective of the bit need of the corresponding signal block of at least the second subband signal component. In aforementioned Dutch Patent Application, document (6a), there has been described that in the bit allocation step bits may be allocated to a signal block in advance. This achieves that this signal block containing at least the number of bits allocated thereto in the initial bit allocation step can be coded with certainty. It has not been described how, when coding, for example, a stereo signal in which first and second subband signal components are located in a subband, the initial bit allocation for the two subband signal components may be realised. Said Dutch Patent Application only describes for a single subband signal the manner in which it can be determined whether or not initial bit allocation is necessary on the basis of the power v.sub.i (t) of a signal block, the masked quantizing power w.sub.i (t) of the signal block and the bit allocation procedure for previous signal blocks of the subband signal.

According to the invention, interaction is now realised between determining initial bit allocation for corresponding signal blocks in a subband, in the way that if there is a determination that initial bit allocation is necessary for one of the corresponding signal blocks, bits are also initially allocated to at least a second corresponding signal block. One preferably allocates in advance equally many bits to the two or more signal blocks. Experiments have shown that such an interaction provides improved auditory perception of the encoded signals.

In addition, if there is a possibility of establishing whether no bits whatsoever need be allocated to a subband signal during bit allocation, there are in fact three options. In this situation, the determining means may now determine that (a) bits are to be allocated initially, or (b) bits are not to be allocated initially, or (c) no bits at all are to be allocated during the bit allocation procedure. The invention as described hereinbefore thus assumes that if the situation (a) occurs for at least one of the subband signal components, bits are to be allocated in advance to at least two subband signal components. This is to say, that the main concept is based on the supposition that there is at least interaction for the case where the situation (a) occurs for one of the corresponding subband signal components in a subband.

This concept of interaction may also be applied for situation (b) as required. Let us assume that for at least two corresponding subband signal components the situation (b) occurs for at least one of the subband signal components, whereas the situation (a) does not occur for any of the signal components, bits will then be allocated to all these subband signal components in the manner described in document (6a).

For the quantization step the first and second subband signal components are put together to form a combined subband signal for subbands to be coded in the intensity mode. There are various ways in which the two subband signal components may be combined. For a description of this combining of the two subband signal components in a subband reference be made to Dutch Patent Application No. 91.00.173 (to which U.S. patent application Ser. No. 07/829,789 corresponds), document (3). Also for this combined subband signal an initial bit allocation may be necessary before the bit allocation is applied to the subband signals. In identical manner to the one stated above for the first and second subband signal components, it may now be determined whether initial bit allocation is necessary for the combined subband signal. Starting from the original first and second subband signal components there is again determined whether an initial bit allocation for at least either of the two subband signal components is necessary. If so, a plurality of bits is allocated in advance to the signal block of the combined subband signal. If it is also possible to determine in advance whether no bits need be allocated to a signal block of a subband signal component, expansion of the concept of interaction to afore-mentioned situations (b) and (c) is necessary so as to be able to determine for the combined subband signal whether an initial bit allocation for the signal block of the combined subband signal is necessary or not, or whether no bits at all will be allocated to the signal block of the combined subband signal.

The invention will be further explained in the description of the drawing figures with reference to a number of exemplary embodiments.

BRIEF DESCRIPTION OF THE DRAWING

FIG. 1 shows the encoding system;

FIG. 1a shows the left and right subband signal components constituted by q-sample signal blocks, plotted against time;

FIG. 2 shows a first embodiment of the determining means;

FIGS. 3, 4 and 5 show the various allocation steps as a function of the power v.sub.i of a signal block of a subband signal component;

FIG. 6 shows a second;

FIG. 7 shows a third embodiment of the determining means; and

FIG. 8 shows a system for coding first and second subband signal components in a subband SB.sub.p in an intensity mode.

DESCRIPTION OF THE PREFERRED EMBODIMENT

FIG. 1 shows an embodiment of the encoding system for coding a stereo audio signal. For example, 16-bit samples of the left signal component of the audio signal are applied to input 1 with a sampling frequency of 44 kHz. The audio signal is applied to a subband splitter 2. The subband splitter 2 splits up the left audio signal component into M subbands by means of a number of M filters, that is to say, a low-pass filter LP, M-2 band-pass filters BP and a high-pass filter HP. For example, M is equal to 32. The sampling frequency of the M left subband signal components is reduced in the blocks referenced 9. In these blocks the sampling frequency is reduced by a factor of M. The signals thus obtained are presented at the outputs 3.1, 3.2, . . . 3.M. At the output 3.1 the signal is presented in the lowest subband SB.sub.1. At the output 3.2 the signal is presented in the lowest but one subband SB.sub.2. At the output 3.M the signal is presented in the highest subband SB.sub.M. The signals at the outputs 3.1 to 3.M have the form of successive samples expressed in 16 or over, for example 24-bit numbers. The samples of the left subband signal component thus appear at the outputs 3.1 to 3.M in FIG. 1. These samples are referenced 1[k].

16-bit samples of the right signal component of the audio signal are presented with a 44 Khz sampling frequency at output 11. The signal is applied to a subband splitter 12 which distributes the right audio signal component over M subbands by means of M filters which are identical with the filters in the splitter 2 as regards their filtering function. Subsequently, the sampling frequency of the M right subband signal components is reduced in the blocks referenced 19. The signals thus obtained are presented at the outputs 13.1 to 13.M. At the output 13.1 there is again available the signal from the lowest subband SB.sub.1 and at the output 13.M the signal from the highest subband SB.sub.M. The signals are also in the form of samples having identical numbers of bits with the signals presented at the outputs 3.1 to 3.M of the splitter 2. The samples are referenced r[k].

FIG. 1a shows the two signal components in each subband plotted against time. The signal stream of the successive samples in the two signal components in each subband are combined to q-sample corresponding time-equivalent signal blocks as is apparent from FIG. 1a. For example, q is equal to 12.

In the present embodiment the subbands SB.sub.1 to SB.sub.M have all identical widths. This is not necessary, however. The prior-art publication (4), Krasner, discusses, for example, a subdivision into a number of subbands whose bandwidths approximately correspond to the bandwidths of the critical bands of the human auditory system in the respective frequency ranges.

The operation of the subband splitters 2 and 12 will not be further discussed, because their operation has already been extensively discussed. To this end reference should be made to prior-art documents (1), (4) and (5) which are considered included in this application where deemed necessary.

The corresponding signal blocks of q successive samples of the left subband signal components are applied to the associated quantizers Q.sub.11 to Q.sub.M1. In a quantizer Q.sub.m1 the samples in a signal block are quantized to quantized samples comprising a number of bits n.sub.m1 smaller than 16.

Similarly, the corresponding signal blocks of the right subband signal components are applied to the associated quantizers Q.sub.1r to Q.sub.Mr. In a quantizer Q.sub.mr the samples in a signal block are quantized to quantized samples comprising a number of bits n.sub.mr smaller than 16.

Prior to quantization, the q samples in a signal block are first normalized. This normalization is effected by dividing the amplitudes of the q samples by the amplitude of the sample having the largest absolute value in the signal block. The amplitude of the sample having the largest amplitude in the signal block provides the scale factor SF, cf. document (2). Subsequently, the amplitudes of the normalized samples, which are now situated in an amplitude range from -1 to +1, are quantized.

In prior-art document (2) this quantization is extensively discussed, cf. FIGS. 24, 25 and 26 and the associated descriptions in that document.

The quantized samples of the left signal components in the subbands SB.sub.1 to SB.sub.M are thereafter presented at the respective outputs 4.1 to 4.M. The quantized samples of the right signal components in the subbands SB.sub.1 to SB.sub.M are presented at the respective outputs 14.1 to 14.M.

The outputs 3.1 to 3.M are further coupled to the respective inputs 5.1 to 5.M of the unit 16.1 belonging to the bit need determining means 6. Furthermore, the outputs 13.1 to 13.M are coupled to the respective inputs 15.1 to 15.M of a unit 16.r belonging to the bit need determining means 6. The units 16.1 and 16.r determine the bit needs b.sub.m1 and b.sub.mr for q-sample signal blocks corresponding with time of the left and right subband signal components in the subbands SB.sub.1 to SB.sub.M. The bit needs b.sub.m1 and b.sub.mr are numbers related to the number of bits with which the q samples in a q-sample signal block of the left and right signal components of a subband-m signal ought to be quantized.

The bit needs b.sub.11 to b.sub.M1, and b.sub.1r to b.sub.Mr derived by the bit need determining means 6 are applied to the bit allocation means 7. The bit allocation means 7 determine, on the basis of the bit needs, the real numbers of bits n.sub.11 to n.sub.M1 and n.sub.1r to n.sub.Mr with which the q samples of the corresponding signal blocks of the left and right subband signal components in the subband signals SB.sub.1 to SB.sub.M are quantized. Control signals corresponding to the numbers n.sub.11 to n.sub.M1 are applied to the respective quantizers Q.sub.11 to Q.sub.M1 over the lines 8.1 to 8.M, so that the quantizers are capable of quantizing the samples of the left signal components with the correct numbers of bits. Control signals corresponding to the numbers of n.sub.1r to n.sub.Mr are applied to associated quantizers Q.sub.1r to Q.sub.Mr, over the lines 18.1 to 18.M, so that also these quantizers are capable of quantizing the samples of the right signal components with the correct numbers of bits.

The documents (6a) and (6b) of the list of references extensively discuss how the bit need determining means 6 and the bit allocation means 7 may function, and characteristics of the samples in a signal block on which the determinations may be based.

The documents (6a) and (6b) explain how the powers v.sub.m1 and the magnitudes w.sub.m1 may be derived from the samples in the corresponding signal blocks of the left subband signal components, and how the bit needs b.sub.m1 may be derived from the magnitudes w.sub.m1 and the scale factors SF.sub.m1. The magnitude w.sub.m1 then represents the power of the masked quantization noise in a signal block of the left subband signal component in the subband SB.sub.m. Similarly, the unit 16.r derives the powers v.sub.mr and the magnitudes w.sub.mr from the samples in the corresponding signal blocks of the right subband signal components, and the bit needs b.sub.mr from the magnitudes w.sub.mr and the scale factors SF.sub.mr. The magnitude w.sub.mr then represents the power of the masked quantization noise in a signal block of the right subband signal component in the subband SB.sub.m.

The documents then describe the bit allocation as performed in the bit allocation means 7. The bit allocation is mainly described for a mono signal. In the described bit allocation algorithm the available number of bits B, starting from the calculated bit needs b.sub.1 to b.sub.M, are distributed over the samples in the corresponding signal blocks in the subbands so as to obtain the magnitudes n.sub.1 to n.sub.M. In the described method always the largest bit need b.sub.i is determined in a number of cyclic steps, after which a number of bits p per sample are allocated to the signal block in the subband SB.sub.i. In the event of bits being allocated to the signal block in the subband SB.sub.i for the first time, p is equal to, for example, 2. If bits are again allocated to the signal block in the subband i at a later stage, p will have a smaller value. For example, p will then be equal to 1.

Above documents also describe that stereo signals may be processed by the bit allocation means 7. In that case there are two options. The first option is as follows.

In this option the bit allocation is separately performed for the left and right subband signal components. In the method discussed previously, the value of B was used for the bit allocation. B was then equal to the number of available bits. It may be evident that in the present case just half the number of available bits are taken for B for the calculation of n.sub.11 to n.sub.M1. The other half of the number of available bits will be used for the bit allocation to the right subband signals for obtaining the values n.sub.1r to n.sub.Mr.

In contradiction of the first option, in which there were separate bit allocations for the left and right subband signals, in the second option the 2M bit needs b.sub.11 to b.sub.M1 and b.sub.1r to b.sub.Mr are applied to a bit allocation unit similar to unit 7. In this unit the 2M numbers of n.sub.11 to n.sub.M1 and n.sub.1r to n.sub.Mr are derived from the real number of available bits B in a manner similar to that described in the two documents with respect to mono signals.

Said documents (6a) and (6b) describe an embodiment of determining means as included in the encoding system. With these determining means there is a possibility that an initial bit allocation to signal blocks may be performed in several subbands. Reference be made to the description of the FIGS. 11 to 14 in said documents. Said documents describe how for successive time intervals, in which corresponding signal blocks of the subbands are coded, it may be determined, on the basis of the powers of v.sub.1 to v.sub.m and the magnitudes w.sub.1 to w.sub.M, that initial bit allocation is to be performed for a signal block or that no initial bit allocation must take place, or that no bits at all need to be allocated to the signal block.

Naturally, the described method could be applied to each separate subband signal component to be thus in a position to establish for the stereo signal whether an initial bit allocation is necessary or not and whether perhaps no bits at all need to be allocated to a signal block.

FIG. 2 shows an embodiment of the determining means according to the invention. The embodiment comprises two sections 20.1 and 20.2 denoted by dashed lines. The two sections are identical. A section, for example, section 20.1 has already been described and shown in above documents, cf. the description of FIG. 11 in these documents. A description of the operation of the section 20.1 will be given hereinbelow. This description will be based on the situations represented in the FIGS. 3, 4 and 5 of the present application.

FIGS. 3, 4 and 5 show the situations for the successive signal blocks of the subband signal component in a subband i to which bits are allocated in advance or not to the subband signal component. The drawing Figures show the successive time intervals .DELTA.T in which corresponding signal blocks of the M subbands are processed. In each time interval the power v.sub.i (t) and the magnitude w.sub.i (t) are determined for each subband signal component in each subband SB.sub.i. For the first subband signal component in the subband SB.sub.i the magnitudes v.sub.i1 and w.sub.i2 are computed. If v.sub.i1 (t) exceeds w.sub.i1 (t), bits are allocated in advance to the first subband signal component in the subband SB.sub.i. As is apparent from FIG. 3, this holds for the time intervals prior to t=t.sub.1. FIG. 2 shows in section 20.1 a circuit by which control signals can be derived from the magnitudes v.sub.i1 and w.sub.i1, which signals denote that initial bit allocation is to be performed for the first subband signal component in which case the output of the SR flip-flop 140 is either "high" or "logic 1", or that no bit allocation is performed in which case the output of the SR flip-flop 141 is "high", or that no initial bit allocation is performed in which case the output of the counter 142 is "high". In the latter case, bits may be allocated indeed to the first subband signal component, but this allocation is then performed at a later stage, i.e. in the block 54 and perhaps also in the block 56 according to the method shown in FIG. 5 of above documents.

At the instant t=t.sub.1, v.sub.i1 (t) becomes smaller than w.sub.i1 (t). The output 144 of the comparator 143 now becomes "low", whereas the output 145 of this comparator becomes "high". This "high" signal is applied through the OR gate 147 to the AND gate 148, so that clock pulses are passed to the AND gate 149 with a frequency f equal to 1/.DELTA.T. Since a "high" signal i