KR20010112434A - Using gain-adaptive quantization and non-uniform symbol lengths for audio coding - Google Patents

Abstract

Techniques such as Hoffman coding can be used to represent digital audio signal components more efficiently using non-uniform Gill symbols, which can be represented by other coding techniques using uniform length symbols. Unfortunately, the coding efficiency that can be achieved by Hoffman coding depends on the probability density function of the information to be coded and the Hoffman coding process itself requires considerable processing and memory resources. The coding process using gain-adaptive quantization according to the present invention can realize the advantages of non-uniform length symbols while overcoming the drawbacks of Hoffman coding. In gain-adaptive quantization, the magnitudes of the signal components to be encoded are compared to one or more thresholds and placed in classes according to the result of the comparison. The magnitudes of the components located in one of the classes are modified according to the gain factor related to the threshold used to classify the components. Preferably, the gain factor can be expressed as a function of only the threshold. Gain-adaptive quantization can be used to encode frequency subband signals in split-band audio coding systems. Additional features are disclosed that include cascade gain-adaptive quantization, inner-frame coding, split-spacing and non-overloading quantizers.

Description

USING GAIN-ADAPTIVE QUANTIZATION AND NON-UNIFORM SYMBOL LENGTHS FOR AUDIO CODING}

It is a continuing interest to encode digital audio signals in the form of imposing low information capacity requirements on storage media and transmission channels capable of delivering encoded audio signals at high levels of original quality. Perceptual coding systems use a process of encoding and quantizing audio signals in a manner that uses larger spectral components in the audio signal, making it impossible to mask or hear synthetic quantization noise to achieve this contradictory purpose. In general, it is advantageous to control the shape and amplitude of the quantization noise spectrum so that it is below the psychoacoustic masking threshold of the signal to be encoded.

The perceptual encoding process applies a bank of analysis filters to an audio signal to obtain subband signals having a bandwidth corresponding to a critical band of the human auditory system, and applies the perceptual model to subband signals or some other range of audio signal spectrum. Apply to content to estimate the masking threshold of the audio signal, set a quantization step size for quantizing the subband signal small enough so that the resulting quantization noise is below the estimated masking threshold of the audio signal, and It will be executed by a so-called split-band encoder that quantizes the sub-ban signals according to the quantization step size and assembles a number of symbols representing the quantized sub-band signals into a coded signal. The complementary perceptual decoding process extracts quantized subband signals from the coded signal and reconstructs quantized subband signals therefrom, obtains a dequantized representation of the quantized subband signals, and is indistinguishable from the original audio signal. It can be implemented by a split-band decoder, which applies a bank of synthesis filters to the dequantized representation to generate an audio signal.

The coding processes of such coding systems often use uniform length symbols to represent quantized signal elements or components within each subband signal. Unfortunately, the use of uniform length symbols imposes higher information capacity than necessary. The necessary information capacity can be reduced by using non-uniform length symbols to represent quantization components in each subband signal.

One technique using non-uniform length symbols is the Huffman coding technique of quantized subband-signal components. In general, Huffman code labels are designed using "trainingsignals" that have been selected to represent signals to be coded in practical use. If the mean probability density function (PDF) of the training signals is very close to the PDF of the coded real signal, and if the PDF is not smooth, Huffman coding can provide a very good coding gain.

If the PDF of the coded real signal is not close to the average PDF of the training signals, Huffman coding will not realize the coding gain and will result in a coding penalty that increases the information capacity condition of the coded signal. This problem can be minimized by using multiple codebooks corresponding to different signal PDFs; However, additional storage space is needed to store codebooks and additional processing is required to encode the signal according to each codebook and then select the one that provides the best results.

A nonuniform length symbol in each subband signal that does not depend on the PDF of any feature component values may be used to represent a block of quantized subband-signal components and may be efficiently executed using minimal computation and memory resources. There is a need for technology.

The present invention relates generally to encoded and decoded signals. The present invention will be advantageously used for split-band encoding and decoding in which frequency-subband signals are arbitrarily coded. In particular, the present invention is useful in perceptual audio coding systems.

1 is a block diagram of a split-band coder employing gain-adaptive quantization.

2 is a block diagram of a split-band decoder employing gain-adaptive dequantization.

3 is a flow chart illustrating the steps of an iterative bit-allocation process.

4 and 5 are graphical representations of the results of applying virtual blocks and gain of subband signal components to the components.

6 is a block diagram of a cascaded gain stage for gain-adaptive quantization.

7 and 8 are graphical views of quantization functions.

9A-9C illustrate how a split-spacing quantization function can be implemented using a mapping transform.

10-12 are graphical views of quantization functions.

13 is a block diagram of an apparatus that can be used to perform various aspects of the present invention.

It is an object of the present invention to provide advantages that can be realized by using non-uniform length symbols to represent quantized signal components, such as the subband-signal components of each frequency subband, in a split-band coding system.

The present invention achieves this goal by using a technique that does not depend on the specific PDF of the component values to achieve good coding gain and can be efficiently executed using minimal computation and memory resources. In some applications, coding systems will use the features of the present invention in connection with other techniques such as Huffman coding.

In accordance with the teachings of one aspect of the present invention, a method of coding an input signal includes receiving an input signal and generating a subband-signal block of subband-signal components representing frequency subbands of the input signal; Comparing magnitudes of the components of the subband-signal block with a threshold, placing each component in one of two or more classes according to the component size, and obtaining a gain factor; Applying the gain factor to components located in one of the classes to modify the magnitudes of certain components of the subband-signal block; Quantizing the components of the subband-signal block; And assembling non-uniform length symbols representing quantized subband-signal components and control information transmitting the classification of the components into a coded signal.

In accordance with the teachings of another aspect of the present invention, a method for decoding an encoded signal receives an encoded signal and obtains control information and non-uniform length symbols therefrom, and frequency subbands of the input signal from the non-uniform length symbols. Obtaining quantized subband-signal components indicative of a; Dequantizing the subband-signal components to obtain subband-signal dequantization components; Applying a gain factor to modify the magnitudes of predetermined dequantization components in accordance with the control information; And generating an output signal in response to the subband-signal dequantization components.

Such methods will be embodied in the medium as program instructions that may be executed by an apparatus for carrying out the invention.

In accordance with the teachings of another aspect of the present invention, an apparatus for encoding an input signal comprises a subband-signal block of subband-signal components having an input for receiving the input signal and representing a frequency subband of the input signal. An analysis filter having an output provided; A subband-signal coupled to the analysis filter that compares the magnitudes of the components of the subband-signal block with a threshold, and places each component in one of two or more classes according to the component size, and obtains a gain factor Block analyzer; A subband-signal component processor coupled to the subband-signal block analyzer for applying the gain factor to the component located in one of the classes to modify the magnitudes of predetermined components of the subband-signal block; A first quantizer coupled to the subband-signal processor for quantizing the components of the subband-signal block having modified magnitudes according to the gain factor; And a formatter coupled to the first quantizer for assembling non-uniform length symbols representing the quantized subband-signal components and control information transmitting the classification of the components into a coded signal.

In accordance with the teachings of another aspect of the present invention, an apparatus for decoding an encoded signal, the apparatus receives the encoded signal and obtains control information and non-uniform length symbols therefrom, and the non-uniform length symbols A deformatter for obtaining quantized subband-signal components from the quantized subband-signal length symbols; A first dequantizer coupled to the deformatter for dequantizing certain subband-signal components of the block according to the control information to obtain first dequantization components; A subband-signal block processor coupled to the first dequantizer for applying a gain factor to modify magnitudes of predetermined first dequantization components of the subband-signal block in accordance with the control information; And a synthesis filter having an input coupled to the subband-signal processor and having an output to which an output signal is provided.

According to another aspect of the teachings of the present invention, a medium may comprise: (1) non-uniform length symbols representing quantized subband-signal components corresponding to an element of a subband signal block representing a frequency subband of an audio signal; (2) control information indicating a classification of the quantized subband-signal components according to sizes of the corresponding subband-signal block element; And (3) an indication of a gain factor suitable for the magnitudes of the quantized subband-signal components in accordance with the control information.

Various features of the present invention and its preferred embodiments will be better understood by referring to the accompanying drawings in which the following description and similar reference numerals refer to similar elements in the several views. The contents and drawings in the following description are set forth by way of example only and should not be understood as limiting the scope of the invention.

A. Coding System

The present invention proceeds to improve the efficiency of representing quantized information such as audio information and seeks to benefit from coding systems using split-band encoders and split-band decoders. Embodiments of a split-band encoder and a split-band decoder employing various aspects of the present invention are shown in FIGS. 1 and 2, respectively.

1. Encoder

a) analysis filtering

In FIG. 1, analysis filterbank 12 receives an input signal from path 11, divides the input signal into subband signals representing frequency subbands of the input signal, and divides the subband signals into path 13. And 23). For the sake of clarity, the embodiments shown in FIGS. 1 and 2 only show components for two subbands; However, in general, the split-band encoder and decoder of a perceptual coding system can handle numerous subbands with bandwidths comparable to the threshold bandwidths of a human auditory system.

Analysis filterbank 12 includes a variety of time-domain to frequency-domain transforms, including cosine-modulated filterbank transforms and wavelet transforms, including polyphase filters, lattice filters, quadrature mirror filters (QMFs), and Fourier-series transforms. Will be implemented in a variety of ways. In a preferred embodiment, the bank of filters is implemented by weighting or modulating blocks of overlapped digital audio samples with an analysis window function and applying a specific modified discrete cosine transform (MDCT) to the window-weighted blocks. These MDCTs are cited as time-domain aliasing cancellation (TDAC) variants and are published in the May 1987 International Conference on Acoustics, Tones and Signals, Princen, Johnson and Bradley (pp.2161-2164). Bradley's article "Subband / Transform Coding Using Filter Bank Designs Based on Time Domain Aliasing Cancellation". Although the scope of implementation can have a profound effect on the performance of the coding system, any particular implementation of the analysis filterbank is not critical to the concept of the present invention.

Each of the subband signals passed along paths 13 and 23 includes subband-signal components arranged in a block. In a preferred embodiment, each subband-signal block is represented in block-scale form in which the components are in proportion to the scale factor. For example, a block-floating-point (BFP) form will be used.

For example, if analysis filterbank 12 is implemented with a block transform, apply the transform to a block of input signal samples to generate a block of transform coefficients, and then 1 to form subband-signal blocks. Subband signals are generated by grouping one or more adjacent transform coefficients. For example, if analysis filterbank 12 is implemented by another type of digital filter, such as QMF, the filter may be passed through a series of input signal samples to generate a series of subband-signal samples for each frequency subband. And then subband signals are generated by grouping the subband-signal samples into blocks. The subband-signal components for these two examples are transform coefficients and subband-signal samples, respectively.

b) perceptual modeling

In a preferred embodiment for perceptual coding modeling, the encoder sets an individual quantization step size to quantize each subband signal using the perceptual model. One way of using the perceptual model to allocate bits properly is shown in FIG. 3. According to this method, step 51 applies a perceptual model to information representing the characteristics of the input signal to set a predetermined quantization noise spectrum. In many embodiments, the noise levels in this spectrum correspond to the estimated psychocore masking threshold of the input signal. Step 52 sets an initially presented quantization step size to quantize the components of the subband-signal block. Step 53 determines the allocation bits needed to obtain the suggested quantization step size for all subband-signal components. Advantageously, a permit is made for the noise-diffusion effect of the synthesis filterbank of the split-band decoder used to decode the coded signal. Several methods of making such a permit are described in U.S. Patent Application No. 09 / 289,865 to Ubale et al., U.S. Pat. All are incorporated into the text by reference.

Step 54 determines whether the total number of allocations required is significantly different from the total number of bits available for quantization. If the total allocation is too high, step 55 increases the presented quantization step size. If the total allocation is too low, step 55 reduces the presented quantization step size. The process returns to step 53 and repeats this process until step 54 determines that the total allocation needed to obtain the proposed quantization step size is close enough to the total number of available bits. Thereafter, step 56 quantizes the subband-signal components according to the set quantization step size.

c) gain-adaptive quantization

Gain-adaptive quantization can be employed in the method described above, for example, by including various aspects of the present invention in step 53. The method described above is typical of many perceptual coding systems, but is only one example of a coding process that may be employed in the present invention. The present invention is in essence used in coding systems that use subjective and / or objective criteria to establish a step size for quantizing signal components. In order to facilitate the discussion, simple embodiments are used in the text to illustrate various aspects of the present invention.

The subband-signal block for one frequency subband is passed along the path 13 to the subband-signal analyzer 14, which measures the magnitude and threshold of the subband-signal components in each block. Are compared and each component is placed in one of two classes depending on the component size. Control information conveying classes of components is passed to the formatter 19. In a preferred embodiment, components with magnitudes below the threshold are placed in the first class. The subband-signal analyzer 14 also obtains a gain factor for next use. As described below, the value of the gain factor is preferably related to the level of the threshold in some way. For example, the threshold may be expressed as a function of only the gain factor. Alternatively, the threshold may be expressed as a function of gain factor and other considerations.

Subband-signal components located in the first class are passed to a gain element 15, which provides the gain coefficients obtained by the subband-signal analyzer 14 to each component of the first class, gain-straining. Components are then passed to quantizer 17. Quantizer 17 quantizes the gain-modified components according to the first quantization step size and passes the resulting quantization components to formatter 19. In a preferred embodiment, the first quantization step size is set according to the perceptual model and the value of the threshold obtained by the subband-signal analyzer 14.

Subband-signal components not located in the first class are passed along the path 16 to the quantizer 18, which quantizes these components according to the second quantization step size. The second quantization step size will be the same between the first quantization steps; However, in a preferred embodiment, the first quantization step size is smaller than the first quantization step size.

The subband-signal block for the second frequency subband is passed along the path 23 and is subbanded by the subband-signal analyzer 24, the gain element 25, and the quantizers 27 and 28. It is treated in the same manner as described above for the band. In a preferred embodiment, the threshold used for each frequency subband is suitable and independent of the threshold used for other frequency subband frequencies.

d) coded signal formatting

The formatter 19 assembles the non-uniform length symbols representing the quantized subband-signal components into control signals that transmit the classification of the components into a coded signal and baseband or modulated communication paths through the spectrum including ultrasonic to ultraviolet frequencies. Pass the encoded signal along a path 20 delivered by a transmission medium comprising or a storage medium including a magnetic tape, a magnetic disk, and an optical disk to convey information using magnetic or optical recording techniques.

The symbols used to represent the quantization components may be the same as the quantization values or the type of code derived from the quantization values. For example, the symbols may be obtained directly from a quantizer or by some processing such as Huffman encoding the quantization values. The quantization values themselves can easily be used as non-uniform length symbols because the number of non-uniform bits can be assigned to the quantized subband signal components of the subband.

2. Decoder

a) Decoded coded signal

In FIG. 2, the deformatter 32 receives symbols from the path 31 and obtains symbols representing control information from which the quantized subband-signal components and the classification of the components are transmitted. Decoding processes can be applied to derive quantization components from the symbols as needed. In a preferred embodiment, the gain-modified components are located in the first class. Deformatter 32 also obtains any information that may be needed, for example, by any perceptual model or bit allocation processes.

b) gain-adaptive dequantization

Dequantizer 33 receives the components for one subband-signal block located in the first class, dequantizes them according to the first quantization step size, and passes the result to gain element 35. do. In a preferred embodiment, the first quantization step size is set according to the threshold that was used to classify the perceptual model and the subband-signal components.

Gain element 35 applies a gain factor to the dequantization components received from dequantizer 33 and passes the gain-modified components to merge 37. The operation of the gain element 35 reverses the gain variation provided by the gain element 15 of the companion encoder. As mentioned above, this gain factor is preferably related to the threshold that was used to classify the subband-signal components.

Subband-signal components not located in the first class are passed to dequantizer 34, which dequantizes those components according to the second quantization step size, and passes the results to integrator 37. The second quantization step size may be equal to the first quantization step size; However, in a preferred embodiment, the second quantization step size is smaller than the first quantization step size.

The integrator 37 forms a subband-signal block by merging the gain-modified dequantization components received from the gain element 35 with the dequantization components received from the dequantizer 36 and the resulting Pass the subband-signal block along path 38 to synthesis filter bank 39.

The quantization components of the subband-signal block for the second frequency subband are described above for the first frequency subband by dequantizers 43 and 44, gain element 45 and integrator 47. And the resulting subband-signal block is passed along the path 48 to the synthesis filterbank 39.

c) composite filtering

Synthesis filterbank 39 may be implemented in a variety of ways complementary to those described above for implementing block filterbank 12. The output signal is generated along path 40 in response to blocks of subband-signal components received from paths 38 and 48.

B. Features

1. Subband Signal Classification

a) brief subband threshold function

The effect of gain-adaptive quantization can be evaluated by referring to FIG. 4, which shows the virtual blocks 111, 112 and 113 of subband-signal components. In the example shown, each subband-signal block consists of eight components numbered from one to eight. Each component is represented by a vertical line, and the size of each component is represented by the height of each line. For example, component 1 of block 111 has a size slightly greater than the value 0.25 as shown on the ordinate axis of the graph.

Line 102 represents the threshold at the 0.50 level. Each component of block 111 will be placed in one of two classes by comparing each component size with the threshold. Components with magnitudes below the threshold are located in the first class. The remaining components are located in the second class. Alternatively, if the components are classified by placing those components in the first class that have an appropriately smaller magnitude than the threshold, slightly different results will be obtained. To facilitate the discussion, a threshold comparison made according to the first example will be assumed and will be discussed in more detail in the text.

The components of block 112 are obtained by applying a gain factor to each block 111 component located in the first class. For example, the size of component 1 of block 112 is slightly larger than 0.500 and is obtained by multiplying the size of component 1 of block 111 by such a gain factor. Conversely, the size of component 2 in block 112 is equal to the size of component 2 in block 111 because these components are located in the second class and are not deformed by the gain factor.

Line 104 represents the threshold at 0.25 level. Each component of block 111 will be located in one of two classes by comparing each component size with this threshold and placing components having a size below the threshold in the first class. The remaining components are located in the second class.

The components of block 113 are obtained by applying a gain factor 4 to each block 111 component located in the first class. For example, the size of component 3 of block 113 is about 0.44, which is obtained by multiplying the size of component 3 of block 111, about 0.11, by a gain factor of 4. Conversely, the size of component 1 in block 113 is equal to the size of component 1 in block 111 because these components are located in the second class and are not modified for the gain factor.

The threshold will be expressed as a function of the gain factor only. As shown by these two examples, the threshold will be expressed as follows.

(One)

Th = threshold; And

G = gain factor.

b) transformation threshold function

Unfortunately, the threshold obtained from Equation 1 is too large because the subband-signal component with a size slightly smaller than the threshold Th overloads the quantizer when transformed by the gain factor G.

The value is called to overload the quantizer if the quantization error of that value exceeds half of the quantization step size. For a symmetric quantizer with a uniform quantization step size that quantizes values in the range of approximately -1 to +1, the positive amount of region overloading the quantizer will be expressed as

(2a)

The range of negative values that overload the quantizer will be expressed as

(2b)

Q _OL = value that overloads the quantizer;

Q _MAX = maximum positive quantization value; And

Q = quantization step size.

For a b-bit symmetric mid-tread code quantizer with a uniform quantization step size that quantizes values in the range of approximately -1 to +1, the maximum positive quantization value (Q _MAX ) is 1-2 ^1-b , and the quantization step size ΔQ is 2 ^1-b ; Half of the quantization step size is 2- ^b . Equation 2a for positive overload values can be rewritten as

(3a)

Equation 2b for a negative overload value can be rewritten as

(3b)

Line 100 in FIG. 4 represents the boundary of the positive overload value for the 3-bit symmetric center-tread code quantizer. The negative range of this quantizer is not shown. (-) a maximum amount of such groups quantizing quantized values of is 0.75 = (1-2 ^1-3) is a half of the quantization step size is 0.125 = 2 ^-3; Thus, for this quantizer the boundary for positive overload values is 0.875 = (1-2 ^-3 ). The boundary for negative overload values is -0.875.

Component 5 of block 111 has a size slightly less than the threshold at value 0.500. When gain factor 2 is applied to this component, the magnitude of the result exceeds the overload boundary of the quantizer. A similar problem results for component 6 when threshold 0.25 is used with gain factor 4.

The threshold for the amount of positive that avoids overload and optimally maps the region of positive component values to the positive range of positive in the first class is expressed as follows.

(4a)

The threshold for the negative amount is expressed as

(4b)

Throughout the remainder of this discussion, only positive thresholds will be discussed. This simplification does not lose any generality, since such an operation of comparing the component sizes with a positive threshold is the same as any other operation that compares the component sizes with a positive and negative threshold.

For the b-bit symmetric center-tread code quantizer described above, the threshold function of equation 4a can be rewritten as follows.

(5)

The effect of gain-adaptive quantization using this variation threshold is shown in FIG. 5, which illustrates the virtual blocks 121, 122, 123 and 124 of the subband signal components. In the example shown, each subband signal block consists of eight components numbered from 1 to 8, the magnitudes of which are represented by the length of each vertical line. Lines 102 and 104 represent thresholds for the 3-bit symmetric center tread code quantizer for gain coefficients such as 2 and 4, respectively. Line 100 represents the boundary of the positive overload values of this quantizer.

The components of the subband-signal block 122 will be obtained by comparing the threshold 102 with the magnitude of the components of the block 121 and applying gain G = 2 to the components having a magnitude below the threshold. Similarly, the components of subband-signal block 123 are obtained by comparing the threshold 104 with the magnitude of the components of block 102 and applying a gain G = 4 to the components having magnitudes below this threshold. Will be. The components of the subband-signal block 124 will be obtained using the cascaded technique described below. Unlike the examples shown in FIG. 4, for the first threshold described above, none of the gain-modified components shown in FIG. 5 exceed the overload boundary of the quantizer.

On the one hand, the deformation threshold according to equation 5 is preferred because it avoids quantizer overload for small-size components of the first class and loads the quantizer appropriately. On the other hand, such a threshold is undesirable in some embodiments looking for an optimal quantization step size, because the threshold cannot be determined until the quantization step size is set. In an embodiment in which quantization step size is adapted by allocating bits, the quantization step size cannot be set until the bit allocation for each subband signal block is known. These disadvantages are described in more detail below.

2. Quantization

Preferably, the quantization step size of the quantizer used to quantize the components of the subband-signal block is adapted in response to the gain factor for that block. In one embodiment using a process similar to that described above and shown in FIG. 3, the number of bits (b) is assigned to each component in the subband-signal block, and then the quantization step size and maybe bit allocation is selected for that block. The gain factor makes it suitable for each component. For this embodiment, the gain factor is selected from four possible values representing gains of 1, 2, 4 and 8. The components of the block are quantized using a symmetric center-tread code quantizer.

Larger-size components that are not located in the first class and whose gain is not modified assign the same number of bits (b) to be allocated, regardless of the benefits of the present invention. In a variant embodiment using the division-interval quantization function discussed below, the bit allocation for these larger-sized components may be reduced for some gain coefficients.

The smaller-sized components located in the first class and whose gain is modified assign the number of bits according to the values shown in Table I.

benefit Assignment One b 2 b-1 4 b-2 8 b-3

Indicates that for a particular subband-signal block a gain factor of 1 does not apply to that block; Thus, an equal number of bits (b) is assigned to each component to be assigned regardless of the benefit of the present invention. The use of gain coefficients G = 2, 4, and 8 for a particular subband-signal block potentially reduces the allocation of 1, 2, and 3 bits, respectively, potentially for each smaller-size component of the susband block. Can be provided as

The allocation shown in Table I applies to the limitation that the number of bits allocated to each component cannot be less than one. For example, if a bit-allocation process that assigns b = 3 to components of a particular subband-signal block and a gain factor G = 8 is selected for that block, then the bit allocation for smaller-sized components is As suggested by Table I, it will be reduced to one bit rather than zero bit. The intended effect of gain modification and adjustment on bit allocation is to use the same number of bits to maintain the same signal-balanced noise ratio. If desired, the embodiment will avoid selecting a gain factor that does not reduce the number of allocated bits.

3. Control Information

As noted above, the subband-signal analyzer 14 provides control information to the formatter 19 for assembling into the coded signal. This control information carries a classification for each component of the subband-signal block. Such control information may be included in the coded signal in various ways.

One way to include the control information is to have the coded signal include a string of bits for each subband-signal block in which one bit corresponds to each component in the block. A bit set to one value, e.g. a value of 1, indicates that the corresponding component is not a gain modified component, and for example, a bit set to another value of value 0 means that the corresponding component is a gain modified component. To indicate. Another way to include the gain information should include a special " escape code " in the coded signal immediately before each component that is gain modified or otherwise not gain modified.

In the above preferred embodiment using a symmetric mid-tread signed quantizer, each large-size component that is not gain modified is run by an escape code equivalent to an unused quantization value. . For example, the quantization values for a 3-bit two's complement coded quantizer range from a minimum of -0,750 represented by a three bit binary string b'101 to a maximum of +0.75 represented by a binary string b'011. The binary string b'100 corresponding to -1.000 is not used for quantization but is valid for use as control information. Similarly, the unused binary string for a 4-bit two's complement coded quantizer is b'1000.

Referring to the subband-signal block 121 of FIG. 5, the components 4, 5 are large-scale components that exceed the threshold 102. As shown in FIG. If this threshold is used in conjunction with gain factor G = 2, then the bit allocation for all small-size components located in the first class is b-1 indicated in Table 1 above. If the bit-allocation process assigns b = 4 bits to each component of block 121, for example, the allocation for each subband-signal component is reduced to 3 = (b-1) bits and 3 A -bit quantizer is used to quantize small-size components. Each large-size component, components 4 and 5 in this embodiment, is quantized with a 4-bit quantizer and identified by the same control information as the unused binary string of the 3-bit quantizer, or b'100 . This control information for each large-size component can be conveniently assembled into the coded signal immediately before each large-size component.

It may be taught that the present invention does not provide any benefit in the embodiments discussed in the foregoing paragraphs. In this embodiment the cost or overhead required to transmit 6 bits of control information is equal to the number of bits saved by reducing the bit allocation for small-size components. Referring to the above embodiment, if only one component of block 121 is a large-size component, the present invention reduces the number of bits required to transmit such a block by four. Seven bits are reduced by reduced allocations to seven small-size components and only three bits are required to transmit control information for one large-size component.

This last embodiment ignores one additional aspect. Two bits are required for each subband-signal block of this embodiment to transmit what four gain coefficients are used for that block. As noted above, a gain factor equivalent to 1 can be used to indicate that features of the present invention are not applied for a particular subband-signal block.

The present invention does not provide any advantage for quantizing subband-signal blocks having four or less components. In perceptual coding systems that generate subband signals having a bandwidth corresponding to the threshold bandwidth of the human auditory organ, the number of components of the subband-signal blocks for low-frequency subbands is probably one component per block. Although low alone, the number of components per subband-signal block is increased by increasing the subband frequency. As a result, in a preferred embodiment, the processing required for implementation features of the present invention may be limited for wider subbands. Additional pieces of control information may be included in the coded signal to indicate the lowest frequency subband in which gain-adaptive quantization is used. The encoder can appropriately select this subband according to the input signal characteristics. This technique avoids the need to provide control information for subbands that do not use gain-adaptive quantization.

4.Decoder Features

Decoders employing features of the present invention may suitably change the quantization step size of the dequantizer in essentially any manner. For example, a decoder devised to decode the coded signal generated by the embodiments described above may use adaptive bit allocation to set the quantization step size. The decoder can operate in a so-called forward-adaptive system, in which bit allocations can be obtained directly from the coded signal, which is obtained by repeating the same allocation process in which the bit allocations were used for the encoder. It can work in a so-called backward-adaptive system, or it can work in the synthesis of two systems. Allocation values obtained in this way are termed "normal" bit allocations.

The decoder obtains control information from the coded signal to identify the gain coefficients and the classification of the components of each subband-signal block. Continuing the above embodiment, the control information storing the gain factor G = 1 indicates that the gain-adaptive feature is not used and a typical bit allocation b dequantizes the components of that particular subband-signal block. Should be used to For other gain coefficient values, a typical bit allocation value b for the block is used to determine the value of the control information identifying the " escape code " or large-size components. In the above embodiment, the allocation value of b = 4 with the gain factor G = 2 indicates that the control information is a binary string b'100 having a length of 3 = (b-1) bits. The presence of such control information in the coded signal indicates just below large-size components.

The bit allocation for each gain-corrected component is shown in Table 1 and adjusted as described above. Dequantization is performed using an appropriate quantization step size and the gain-corrected components are subject to a gain factor which is the inverse of the gain factor used to perform gain correction at the encoder. For example, if the small-size components are multiplied by the gain coefficient G = 2 of the encoder, the decoder applies the inverse gain G = 0.5 to the corresponding dequantization components.

C. Additional Features

In addition to the above variations, a number of variations are described below.

1. Additional Classifications

According to one variant, the magnitudes of the components of the subband-signal block are compared to two or more thresholds and located in two or more classes. Referring to FIG. 5, for example, the magnitude of each component of block 121 may be compared to thresholds 102 and 104 and may be located in one of three classes. Gain coefficients are obtained for the two classes and applied to the appropriate components. For example, gain factor G = 4 may be applied to components having a magnitude equal to or less than threshold 102 but greater than threshold 104. Alternatively, gain factor G = 2 may be applied to all components having magnitudes equal to or less than threshold 102 and gain factor G = 2 equal to or less than threshold 104. It can be applied again for the components having these.

2.Cascaded Operation

The gain correction process is performed many times before quantization. 6 is a block diagram illustrating an embodiment of one of the two gain stages of the cascade. In this embodiment, the subband-signal analyzer 61 compares the magnitudes of the components of the subband-signal block with a first threshold and places the components in one of two classes. Gain element 62 applies a first gain factor to components located in one of the classes. The value of the first gain factor is associated with the first threshold.

Subband-signal analyzer 64 compares the magnitudes of the gain-corrected components and possibly the remaining components of the block with a second threshold and places the components in one of two classes. Gain element 65 applies the second gain factor to the components located in one of the classes. The value of the second gain factor is associated with a second threshold. If the second threshold is equal to or less than the first threshold, the subband-signal analyzer 64 needs to analyze the components that the analyzer 61 has placed in class for magnitudes above the first threshold. none.

The subband-signal block components are quantized by the quantizers 67 and 68 in a manner similar to that described above.

Referring to FIG. 5, the components of subband-signal block 124 are such that subband-signal analyzer 61 and gain element 62 have a gain factor G = 2 equal to or less than threshold 102. Gain-modified components having a magnitude equal to or less than the threshold 102 where the subband-signal analyzer 62 and the gain element 65 apply to magnitude components. Obtained by successive application of gain stages to For example, components 1-3 and 6-8 of block 121 are modified with a gain factor G = 2 of the first stage, producing an intermediate result shown in block 122. Components 1, 3, 7, and 8 are modified with a gain factor G = 2 in the second stage to obtain the result shown for block 124.

In an embodiment using the gain stages of the cascade, appropriate control information must be provided to the coded signal so that the decoder can perform a complementary set of gain stages of the cascade.

3.Optimization bit allocation

There are a number of possible strategies for applying gain-adaptive quantization. One simple strategy is to analyze the components of the individual subband-signal blocks by starting with a first threshold and associated first gain factor G = 2, and gain-adaptive quantization according to the first threshold and the first gain coefficient. Determines if it causes a reduction in bit allocation requirements. If it is not, the analysis is stopped and no gain-adaptive quantization is performed. If it causes a decrease, the analysis continues with the second threshold and the associated second gain factor G = 4. If the use of the second threshold and associated gain factor does not cause a reduction in bit allocation, gain adaptive quantization is performed using the first threshold and the first gain factor. If the use of the second threshold and the second gain factor causes a decrease, the analysis continues with the third threshold and the associated third gain factor G = 8. This process continues until the use of the threshold and associated gain coefficients does not cause a reduction in bit allocation, or until all combinations of thresholds and associated gain coefficients are considered.

Another strategy allows to optimize the selection of the gain factor by calculating the cost and benefit provided by each possible threshold and associated gain factor and using the threshold and gain factor resulting in maximum net benefit. . For this embodiment, the net benefit for a particular threshold and associated gain factor is total cost minus cost. The gross benefit is the number of bits saved by reducing the bit allocation for small-size components that are gain modified. The cost is the number of bits required to transmit control information for large-size components that are not gain modified.

One way in which this preferred strategy can be implemented is shown in the program section below. This program part is represented in pseudo-code using syntax, which includes some syntactic features of C, FORTRAN, and BASIC programming languages. This program portion and other programs shown herein are not intended to be source code segments suitable for compilation but are provided to transmit some aspects of the possible implementations.

Gain (X, N, b) {

Th2 = (1-2 ^ (-b)) / gf [1]; // initialize threshold for gain factor G = 2

Th4 = Th2 / 2; // initialize threshold for gain factor G = 4

Th8 = Th4 / 2; // initialize threshold for gain factor G = 8

n2 = n4 = n8 = 0; // counter reset

for (k = 1 to N) {// for each component k ..

CompMag = Abs (X [k]); // get component size

if (CompMag> Th2)

n2 = n2 + 1; // Th2 or more components count

else if (CompMag> Th4)

n4 = n4 + 1; // count components between Th4 and Th2

else if (CompMag> Th8)

n8 = n8 + 1; // count components between Th8 and Th4

}

n24 = n2 + n4; // number of components greater than Th4

n248 = n24 + n8; // number of components greater than Th8

benefit2 = Min (b-1,1); // bits per small component saved using G = 2

benefit4 = Min (b-1,2); // bits per small component saved using G = 4

benefit8 = Min (b-1,3); // bits per small component saved using G = 8

net [0] = 0; // net benefit for no gain correction

net profit using net [1] = (N-n2) * benefit2-n2 * (b-benefit2); // G = 2

net profit using net [2] = (N-n24) * benefit4-n24 * (b-benefit4); // G = 4

net profit using net [3] = (N-n248) * benefit8-n248 * (b-benefit8); // G = 8

j = IndexMax (net [j], j = 0 to 3); // get the index of maximum profit

Gain = gf [j]; // get gain factor

}

The function Gain is provided with an array X of subband-signal block components, the number N of components of the block, and a typical bit allocation value b for the block of components. The first statement of the function uses the calculation according to Equation 5 above to initialize the variable Th2 to represent a threshold value related to the gain factor G = 2 obtained from the array gf. In this embodiment, the gain coefficients gf [1], gf [2] and gf [3] are G = 2, 4, 8, respectively. The following statement initializes the variables for the thresholds associated with gain factor G = 4,8. Next, counters are initialized to zero and used to determine the number of large-size components of various classes.

The statement in the for loop calls the function Abs to obtain the magnitude for each subband-signal block component of array X and compares the component size with the thresholds, starting at the best threshold. For example, if the magnitude is greater than or equal to the threshold Th2, the variable n2 is increased by one. When the for loop ends, the variable n2 contains the number of components having a magnitude greater than or equal to the threshold Th2 and the variable n4 includes the number of components having a magnitude greater than or equal to the threshold Th4 but equal to or less than or equal to the threshold Th2. The variable n8 includes the number of components having a magnitude greater than or equal to the threshold Th8 but equal to or less than the threshold Th4.

The two statements of the for loop just below calculate the total number of components that are the individual thresholds. The number of variables n24 indicates the number of components having a magnitude greater than or equal to the threshold Th4, and the number of variables n248 indicates the number of components having a magnitude greater than or equal to the threshold Th8.

The next three statements calculate the benefit for each small-size component using each gain factor. This benefit may be on the order of 1,2 or 3 bits per component as shown in Table 1, but the benefit is also limited to no more than b-1 bits per component since the allocation for each component is limited to a minimum of 1 bit. . For example, the number of variable benefit2 indicates the number of bits per small-size component that is saved using the gain factor G = 2. As shown in Table 1, this benefit may be on the order of one bit, however, the benefit is also limited to not exceeding the usual bit allocation value b-1. The calculation of this benefit is provided using the function Min to return the minimum of two values b-1 and 1.

The net benefits are then calculated and assigned to the elements of the array net. The element net [0] indicates the net benefit of not using zero, gain-adaptive quantization. The net gain using the gain factor G = 2 is multiplied by the appropriate number of small-size components (N-n2) and the appropriate gain per small-size component benefit2, so that the unused quantizer value used for the control information. Is assigned to net [1] by subtracting the cost, which is the number of large-size components n2 multiplied by the length of. This length is the bit-length of the small-size components that can be obtained from a typical bit allocation b reduced by the bits saved per small-size component. For example, when the gain factor G = 2, the bit-length of the small-size components is an amount b-benefit2. Similar calculations are implemented by assigning net gains using gain factors G = 4 and 8 to variables net [2] and net [3], respectively.

The function IndexMax is called to obtain the array index j for the maximum net benefit of the array net. This index is used to obtain the appropriate gain factor from the gf array returned by the function Gain.

4. Improved Efficiency Using Simplified Threshold Functions

It has been mentioned above that various features of the present invention can be incorporated into a perceptual bit allocation process as shown in FIG. In particular, these features can be implemented in step 53. Step 53 is implemented in a loop to repeatedly determine the bit allocations presented for quantizing the components of each subband-signal block to be encoded. For this reason, the efficiency of the operations implemented in step 53 is very important.

The above process for function gain, which determines the optimal gain factor for each block, is relatively inefficient because it must count the number of subband-signal block components located in various classes. The component count must be calculated during each iteration because the thresholds obtained according to equation 5 are not calculated until the bit allocation value b presented for each iteration is known.

In contrast to the threshold values obtained according to equation 5, the threshold values obtained according to equation 1 are less accurate but can be calculated before the presented bit allocation value b is known. This allows the thresholds and component counts to be calculated outside the reiteration. Referring to the method shown in FIG. 3, for example, thresholds Th1, Th2 and Th3, and component counts n2, n24 and n248 can be calculated in step 52.

A variation of the function gain, which can be used in this embodiment, is shown in the program section below.

Gain2 (X, N) {

benefit2 = Min (b-1,1); // bits per small component saved using G = 2

benefit4 = Min (b-1,2); // bits per small component saved using G = 4

benefit8 = Min (b-1,3); // bits per small component saved using G = 8

net [0] = 0; // net benefit for no gain correction

net profit using net [1] = (N-n2) * benefit2-n2 * (b-benefit2); // G = 2

net profit using net [2] = (N-n24) * benefit4-n24 * (b-benefit4); // G = 4

net profit using net [3] = (N-n248) * benefit8-n248 * (b-benefit8); // G = 8

j = IndexMax (net [j], j = 0 to 3); // get the index of maximum profit

Gain = gf [j]; // get gain factor

}

The statements of the function Gain2 are identical to the corresponding declaration of the function Gain, which calculates the net gains for each gain factor to select the optimal gain factor.

5. Quantization Function

a) division interval function

The quantization accuracy of the large-size component can be improved by using a division interval quantization function that quantizes the input value between two non-contact intervals.

Line 105 in FIG. 7 is a diagram of a function illustrating end-to-end effects of a 3-bit symmetric mid-trad code quantum and a complementary quantum. The value on the x-axis represents the input to the quantum and the value on the q (x) axis represents the corresponding output from the dequantizer. Lines 100 and 109 represent the boundaries of positive and negative overload values, respectively, for the quantizer. Lines 102 and 108 represent the positive and negative threshold values for the gain factor G = 2 shown in FIG. Lines 104 and 107 represent positive and negative thresholds, respectively, with a gain factor G = 4.

Referring to FIG. 1, when subband signal analyzer 14 classifies subband signal block components according to threshold 102, it is understood that the magnitudes of the components according to quantizer 18 are all greater than threshold 102. FIG. It is recognized. That is, quantizer 18 may not be used to quantize any value between thresholds 108 and 102. This void indicates under utilization of the quantizer's quantizer.

Under this use can be overcome by using a quantizer that implements a division-interval quantization function. Various split-interval functions are possible. 8 is a diagram of a function illustrating end-to-end effects of one split-spacing 3-bit coded quantizer and a complementary dequantizer. Line 101 represents a function of positive quantities and line 106 represents a function of negative quantities.

The function shown in FIG. 8 has eight levels of quantization as opposed to the function having only seven quantization levels, shown in FIG. Additional quantization levels are obtained using the levels described above for the mid-tread quantization function corresponding to -1.

b) non-overloading quantizer

The 3-bit quantizer and the complementary dequantizer implementing the function shown in Fig. 8 are preferred for quantizing the values within the division-spacing from -1.0 to about-0.5 and +0.5 to +1.0 because the quantizer cannot be overloaded. do. As described above, if the quantization error of the value exceeds the 1/2 quantization step size, the value will overload the quantizer. As can be seen in the example shown in Figure 8, the dequantizer output is limited to the same values as -0.9375, -0.8125, -0.6875, -0.5625, +0.5625, +0.6875, +0.8125 and +0.9375. The magnitude of the dequantization error of all values within the split-interval is not greater than 0.0625, which is equal to the 1/2 quantization step size. Such a quantizer is referred to herein as a "non-overloading quantizer" because it is not overloaded.

Essentially at any quantization step size, non-overloading single- and split-interval quantizers can be realized by implementing a quantization function with quantizer outputs defined by quantizer "decision points" properly spaced within an interval of the value to be quantized. have. In general, decision points are spaced some distance d from the other, and the decision points closest to each end of the input interval are spaced from each end to d. These spaces, when used with complementary dequantizers, provide quantizers that are uniformly spaced apart from each other by a specific quantization step size and that provide uniformly spaced output values with a maximum quantization error of a specific quantization half step size. .

c) mapping function

The division interval quantizer can be implemented in various ways. No particular implementation is important. One implementation shown in FIG. 9A consists of a quantizer 74 and a cascade mapping transform 72. The mapping converter 72 receives input values from the path 71, maps these input values at appropriate intervals, and sends the mapped values along the path 73 to the quantizer 74.

If quantizer 74 is an asymmetric mid-tread code quantizer, the mapping function shown in lines 80 and 81 shown in FIG. 9B is adapted to mapping function 72. According to this mapping function, the value within the interval between -1.0 and -0.5 is linearly mapped from -1.0-½ΔQ to -1/2 ΔQ interval, where ΔQ is the quantization step size of the quantizer 74 and the interval. Values within +0.5 to +1.0 are linearly mapped at intervals of −1 / 2ΔQ to +1.0 −1 / 2ΔQ. In this example, the large-size component also cannot have a value exactly equal to -0.5 or +0.5 since the component having these values is classified as a small-size component. For this reason, the mapping converter 72 does not map any input value exactly to -1 / 2ΔQ; However, any one of -1 / 2ΔQ can be arbitrarily mapped.

The effect of this mapping can be recognized with reference to FIGS. 9B and 9C. Referring to FIG. 9B, the mapping converter 72 may recognize that the input points 82 and 84 are mapped to the mapped points 86 and 88. Referring to Figure 9C, which illustrates a function representing the end-to-end effect of a 3-bit asymmetric mid-tread code quantizer and a complementary dequantizer, the mapped points 86,88 are represented by -1 / 2ΔQ values. It can be seen that it lies on either side of the quantizer decision point 87 with

The complementary split-spacing dequantizer may be implemented by an asymmetric mid-tread code dequantizer complementary to the quantizer 74 accompanied by the mapping transform, which is the inverse of the mapping transform unit 72.

d) complex functions

In the example described above, gain adaptive quantization with gain factor G = 2 is used to quantize the components of the subband signal where the conventional bit allocation value b is equal to 3 bits. As described above in connection with Table I, three bits are used to quantize the large-size component bits, and 2 = (b-1) bits are used to quantize the small-size gain-corrected components. Preferably, a quantizer implementing the quantization function of FIG. 8 is used to quantize the large-size component.

A two-bit symmetric mid-tread code quantizer and complementary dequantizer implementing the function 111 shown in FIG. 10 are used for small-size gain-correction components. The illustrated function 111 takes into account the scaling and descaling effects of the gain factor G = 2 used in connection with the quantizer and the dequantizer, respectively. The outputs of the dequantizer are -0.3333 ..., 0.0 and +0.3333 ... and the quantizer decision points are at -0.1666 ... and +0.1666 ....

Complex functions for large- and small-size components are shown in FIG.

e) Transform split-interval function

The use of a split-spacing quantizer with gain factor G = 2 and a threshold of about 0.500 provides an improvement in quantization resolution of about 1 bit. This enhanced resolution can be used to preserve the quantization resolution of large components while reducing the bit allocation by 1 bit to these components. In Example 2, 2-bit quantizers can be used to quantize large-size and small-size components. The complex of quantization functions implemented by the two quantizers is shown in FIG. Quantizers implementing quantization functions 112 and 113 may be used to quantize large-size components having positive and negative amplitudes respectively and a quantizer implementing quantization function 111 may be used to quantize small-size components. Can be used for

The use of split-spacing quantization functions with larger gain coefficients and smaller thresholds does not provide full bits of improved quantization resolution; Thus, bit allocation cannot be reduced without sacrificing quantization resolution. In a preferred embodiment the bit allocation b for large-size mantissas is reduced by one bit for the gain adaptive quantized block using gain factor G = 2.

The dequantization function provided to the decoder must be complementary to the quantization function used for the encoder.

6.Intra-Frame Coding

The term " encoded signal block " is used herein to refer to encoded information representing all subband-signal blocks for frequency subbands across the useful bandwidth of the input signal. Assemble multiple coded signal blocks into larger units, which are referred to in text as frames of frame structure, which share information across coded signal blocks to reduce information overhead or audio and video signals. Are used in a number of applications to facilitate synchronizing signals such as: A number of issues involved in encoding audio information into frames for audio / video applications are incorporated herein by reference. US Patent No. PCT / US98 / 20751, filed May 17, have.

The feature of the gain-adaptive quantization can be applied to groups of subband-signal blocks in other coded signal blocks. This aspect can be advantageously used, for example, in applications that group coded signal blocks into frames. This technique essentially groups the components of multiple subband-signal blocks into a frame, classifies them and applies a gain factor to the components of this group. This so-called inner-frame coding technique can share information between blocks within a frame. No particular grouping of coded signal blocks is critical to implement this technique.

D. Implementation

The present invention is in a broad manner including software of a general-purpose computer system or any other device that includes more specialized components such as digital signal processor (DSP) circuits coupled to components similar to those included in general purpose computer systems. It can be implemented as. 13 is a block diagram of a device 90 that may be used to implement various aspects of the present invention. DSP 92 provides computing resources. RAM 93 is system RAM. ROM 94 represents some form of permanent storage, such as a read-only memory (ROM) for storing programs necessary to operate device 90 and to perform various aspects of the present invention. I / O controller 95 displays interface circuitry for transmitting and receiving audio signals over communication channel 96. Analog-to-digital converters and digital-to-analog converters may be included in the I / O controller 95 that is required to receive and / or transmit analog audio signals. Connected to a bus 91 capable of displaying; However, bus architectures are not required to implement the present invention.

In embodiments implemented in a general purpose computer system, additional components are included to interface to devices such as a keyboard or mouse and a display, and to control a storage device having a storage medium, such as a magnetic tape or disk, or an optical medium. Can be. The storage medium may include embodiments of programs used to record programs, utilities, and applications of instructions for a computing system and to implement various aspects of the present invention.

The functions necessary for the various various aspects of the present invention may be implemented by components implemented in a variety of ways, including discrete logic components, one or more ASICs, and / or a program-controlled processor. The manner in which these components are implemented is not critical to the invention.

The software implementations of the present invention may use various machine readable media, such as baseband or modulated communication paths over a spectrum including ultrasound to ultraviolet frequencies, or any magnetic or optical recording including magnetic tape, magnetic disks and optical disks. It may be transmitted by a storage medium including those which transmit information using a technology. Various aspects of computer system 90 may be controlled by processing circuits such as ASICs, general-purpose integrated circuits, read-only memory (ROM), or microprocessors controlled by programs included in various forms of RAM, and other techniques. It can be implemented with various components.

Claims (42)

In a method of encoding an input signal, Receiving an input signal and generating a subband-signal block of subband-signal components representing frequency subbands of the input signal; Comparing the magnitudes of the components of the subband-signal block with a threshold, placing each component in one of two or more classes according to the component size, and obtaining a gain factor; Applying the gain factor to the components located in one of the classes to modify the magnitudes of certain components of the subband-signal block; Quantizing the components of the subband-signal block; And Assembling non-uniform length symbols representing the quantized subband-signal components and control information transmitting the classification of the components into a coded signal. Method comprising a. 2. The method of claim 1, wherein the control information is assembled with a coded signal indicating these quantized subband-signal components having magnitudes that are not modified according to the gain coefficients, wherein the control information indicates the quantized subband-signal components. Characterized in that it is transmitted by one or more reserved symbols not used for the purpose. 1. The method of claim 2, comprising retrieving the threshold from a function that is independent of the quantization step size of the quantization components but depends on a gain factor. 3. The method of claim 1 or 2, comprising obtaining the threshold from a function that depends on a gain factor and the quantization step size of the quantization components. The method according to any one of claims 1 to 4, Suitably varying each quantization step size for each component of the subband-signal block according to the class in which the component is located by appropriately assigning bits to the component; Obtaining the gain factor such that the number of bits assigned to components having modified sizes is reduced while preserving the respective quantization step size Method comprising a. 6. The method of any one of claims 1 to 5, comprising quantizing the components located in one of the classes according to a division-interval quantization function. The method according to any one of claims 1 to 6, Place each component in one of three or more classes, depending on the component size, Obtaining one or more additional gain coefficients associated with each class; And Applying each of said additional gain coefficients to said components located in said associated respective class. The method according to any one of claims 1 to 7, Comparing the magnitudes of at least some of the components of the subband-signal block with a second threshold, placing each component in one of the one or more second classes according to the component size, and obtaining a second gain factor; step; And Applying the second gain factor to the components located in one of the second classes to modify the magnitudes of the predetermined components of the subband-signal block, And wherein the non-uniform length symbols represent the quantization components modified by the gain factor and the second gain factor. The method of claim 1, wherein at least some of the components are quantized using one or more non-overloading quantizers. In the method for decoding an encoded signal, Receiving the encoded signal, obtaining control information and non-uniform length symbols therefrom, and obtaining quantized subband-signal components representing frequency subbands of an input signal from the non-uniform length symbols; Dequantizing the subband-signal components to obtain subband-signal dequantization components; Applying a gain factor to modify the magnitudes of certain dequantization components in accordance with the control information; And Generating an output signal in response to the subband-signal dequantization components Method comprising a. 11. The method of claim 10, wherein control information is obtained from the coded signal indicating such quantized subband-signal components having magnitudes that should not be modified in accordance with the gain factor, wherein the control information indicates quantized subband-signal components. Characterized in that it is transmitted by one or more reserved symbols not used. 12. The method of claim 10 or 11, comprising dequantizing the predetermined dequantization components of the subband-signal block in accordance with a dequantization function complementary to a split-interval quantization function. 13. A method as claimed in any of claims 10 to 12, comprising applying a second gain factor to modify predetermined magnitudes of the dequantization components in accordance with the control information. 14. The method of any one of claims 10 to 13, wherein at least some quantization components are dequantized using one or more dequantizers complementary to each non-overloading quantizer. An apparatus for encoding an input signal, An analysis filter having an input for receiving the input signal and having an output provided through a subband signal block of subband signal components representing a frequency subband of the input signal; A subband-signal coupled to the analysis filter that compares the magnitudes of the components of the subband-signal block with a threshold, and places each component in one of two or more classes according to the component size, and obtains a gain factor Block analyzer; A subband-signal component processor coupled to the subband-signal block analyzer that applies the gain factor to the components located in one of the classes to modify the magnitudes of predetermined components of the subband-signal block ; A first quantizer coupled to the subband-signal processor for quantizing the components of the subband-signal block having modified magnitudes according to the gain factor; And A formatter coupled to a first quantizer for assembling non-uniform length symbols representing the quantized subband-signal components and control information transmitting the classification of the components into a coded signal Apparatus comprising a. 16. The apparatus of claim 15, further comprising a second quantizer coupled to the subband-signal block for quantizing the components located in one of the classes in accordance with a division-interval quantization function, wherein the formatter also includes the second quantization. Device coupled to the device. 17. The apparatus of claim 15 or 16, wherein the formatter assembles control information into the coded signal indicating quantized subband signal components having unmodified magnitudes according to the gain factor, wherein the control information is quantized subband-. And is transmitted by one or more spare symbols not used to represent signal components. 18. The apparatus of any one of claims 15 to 17, wherein the threshold is obtained from a function that depends on a gain factor but is independent of the quantization step size of the quantization components. 18. The apparatus of any one of claims 15 to 17, wherein the threshold is obtained from a function that depends on a gain factor and the quantization step size of the quantization components. 20. The method according to any one of claims 15 to 19, wherein the quantization step size for each component of the subband-signal block is appropriately changed and modified according to the class in which the component is located by appropriately assigning bits to the component. And obtain the gain factor such that the number of bits assigned to the components having the specified magnitudes is reduced while preserving the respective quantization step size. 21. The method of any one of claims 15 to 20, wherein each component is placed in one of three or more classes according to component size, and one or more additional gain coefficients associated with each class are obtained, And apply each of said additional gain coefficients to said components located. The method according to any one of claims 15 to 21, The subband-signal block analyzer compares at least some of the components of the subband-signal block with a second threshold, placing each component in one of two or more second classes according to component size, and Obtain two gain factors; The subband-signal component processor applies the second gain factor to the components located in one of the second classes to modify the magnitudes of certain components of the subband-signal block; And the non-uniform length symbols represent the quantization components modified by the gain factor and the second gain factor. 23. The apparatus of any one of claims 15 to 22, wherein at least some of the components are quantized using one or more non-overloading quantizers. An apparatus for decoding an encoded signal, A deformatter for receiving the coded signal to obtain control information and non-uniform length symbols therefrom and to obtain quantized subband-signal components from the non-uniform length symbols; A first dequantizer coupled to the deformatter for dequantizing certain subband-signal components of the block according to the control information to obtain first dequantization components; A subband-signal block processor coupled to the first dequantizer for applying a gain factor to modify magnitudes of predetermined first dequantization components of the subband-signal block in accordance with the control information; And A synthesis filter having an input coupled to the subband-signal processor and having an output provided with an output signal Apparatus comprising a. 25. The second dequantization coupled to the deformatter of claim 24, wherein the second dequantizer coupled to the deformatter dequantizes other subband-signal components of the block according to a dequantization function complementary to the split-interval quantization function to obtain second dequantization components. Apparatus comprising a group. 26. The apparatus of claim 24 or 25, wherein the deformatter obtains control information from the coded signal indicating such quantized subband-signal components having magnitudes that should not be modified in accordance with the gain factor. 27. The apparatus of any of claims 24 to 26, wherein the subband-signal block processor applies a second gain factor to modify a predetermined magnitude of the dequantization components in accordance with the control information. 28. The apparatus of any one of claims 24 to 27, wherein the device quantizes at least some of the quantization components using one or more dequantizers complementary to each non-overloading quantizer. A computer program product comprising program instructions contained on a machine readable medium, the program instructions executable by the machine to implement a method for encoding an input signal, the method comprising: The method, Receiving the input signal and generating a subband-signal block of subband-signal components indicating a frequency subband of the input signal; Comparing the magnitudes of the components of the subband-signal block with a threshold, placing each component in one of two or more classes according to the component size, and obtaining a gain factor; Applying the gain factor to the components located in one of the classes to modify the magnitudes of certain components of the subband-signal block; Quantizing the components of the subband-signal block; And Assembling non-uniform length symbols representing quantum subband-signal components and control information transmitting the classification of components into a coded signal; Computer program product comprising a. 30. The apparatus of claim 29, wherein the control information is assembled into a coded signal indicative of these quantized subband-signal components having magnitudes that are not modified according to the gain factor, the control information representing the quantized subband-signal components. Computer program product, characterized in that it is transmitted by one or more spare symbols not used. 31. The computer program product of claim 29 or 30, comprising obtaining the threshold value from a function that is independent of the quantization step size of the quantization components but depends on a gain factor. 31. The computer program product of claim 29 or 30, comprising obtaining the threshold value from a function that depends on a gain factor and the quantization step size of the quantization components. 33. The method of any of claims 29-32, Appropriately varying each quantization step size for each component of the subband-signal block according to the class in which the component is located by appropriately assigning bits to the component; Obtaining the gain factor such that the number of bits assigned to components having modified sizes is reduced while preserving the respective quantization step size Computer program product comprising a. 35. A computer program product according to any of claims 29 to 34, comprising quantizing the components located in one of the classes in accordance with a partition-interval quantization function. The method according to any one of claims 29 to 34, wherein Place each component in one of three or more classes, depending on the component size, Obtaining one or more additional gain coefficients associated with each class; And And applying respective said additional gain coefficients to said components located in said respective respective classes. The method according to any one of claims 29 to 35, Comparing the magnitudes of at least some of the components of the subband-signal block with a second threshold, placing each component in one of the one or more second classes according to the component size, and obtaining a second gain factor; step; And Applying the second gain factor to the components located in one of the second classes to modify the magnitudes of certain components of the subband-signal block, wherein the non-uniform length symbols And said quantization components modified by a gain coefficient and said second gain coefficient. 37. A computer program product according to any of claims 29 to 36, wherein at least some of the components are quantized using one or more non-overloading quantizers. A computer program product readable by a device comprising a program of instructions for execution by a device to implement a method for decoding an encoded signal, the method comprising: Receiving the coded signal to obtain control information and non-uniform length symbols therefrom, and obtaining quantized subband-signal components representing frequency subbands of an input signal from the non-uniform length symbols; Dequantizing the subband-signal components to obtain subband-signal dequantization components; Applying a gain factor to modify the magnitudes of certain dequantization components in accordance with the control information; And Generating an output signal in response to the subband-signal dequantization components Method comprising a. 39. The apparatus of claim 38, wherein control information is obtained from the coded signal indicating such quantized subband-signal components having magnitudes that should not be modified in accordance with the gain factor, wherein the control information indicates quantized subband-signal components. Computer program product characterized in that it is transmitted by one or more spare symbols not used. 40. The computer program product of claim 38 or 39, comprising dequantizing the predetermined dequantization components of the subband-signal block according to a dequantization function that is complementary to a division-interval quantization function. 41. A computer program product according to any of claims 38 to 40, comprising applying a second gain factor to modify the magnitudes of certain dequantization components in accordance with the control information. 42. The computer program product of any one of claims 38-41, wherein the at least certain quantization components are dequantized using one or more dequantizers complementary to each non-overloading quantizer.