Coding apparatus for digital signal

• US 5,490,170 A
• Filed: 11/30/1993
• Issued: 02/06/1996
• Est. Priority Date: 03/29/1991
• Status: Expired due to Term

• Current Assignee: Ironworks Patents LLC
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Akagiri, Kenzo, Tsutsui, Kyoya
• Primary Examiner(s): Chin, Stephen
• Assistant Examiner(s): Bocure, Tesfaldet
• Application Number:
  US08/159,122
• Time in Patent Office: 798 Days
• Field of Search: 375/122, 375/25, 375/27, 375/240, 375/242, 375/244, 375/245, 381/29, 381/31, 381/30, 381/35, 381/36, 381/37, 348/384, 348/390, 348/398, 348/405, 395/2.13, 395/2.39, 364/725, 364/726
• US Class Current: 375/240
• CPC Class Codes: G10L 19/002 Dynamic bit allocation for perceptual audio coders G10L19/032
• CPC Class Codes: G10L 19/0204 using subband decomposition
• CPC Class Codes: G10L 19/0208 Subband vocoders
• CPC Class Codes: G10L 19/022 Blocking, i.e. grouping of samples in time; Choice of analysis windows; Overlap factoring
• CPC Class Codes: H04B 1/665 using psychoacoustic properties of the ear, e.g. masking effect

Abstract

In a coding apparatus for a digital signal adapted for implementing, every variable length block, floating processing to an input digital signal by using a block floating processing circuit thereafter to orthogonally transform signal components which have undergone such processing by using orthogonal transform circuits (e.g., DCT circuits), the block floating processing circuit is constructed so as to determine the length of a variable length block and a floating coefficient of the block floating processing on the basis of the same index, e.g., a maximum absolute value in that block. Thus, a quantity subject to processing can be reduced. In addition, there may be employed such a configuration to divide, every critical bands, spectrum signals on the frequency base from DCT (Discrete Cosine Transform) circuits to determine, every respective critical bands, allowed noises in which the masking is taken into consideration to compare these allowed noises and a minimum audible curve from a minimum audible curve generator at a comparator. When the minimum audible curve is grater than an allowed noise at that time, this minimum audible curve is considered as an allowed noise to divide the critical band into smaller bands to carry out bit allocation every respective smaller bands, and to rase or set a flag. Thus, an accurate allowed noise level can be provided without increasing auxiliary information.

Claims

1. An apparatus for compressing a digital input signal, the apparatus comprising:

band division filter means for dividing the digital input signal in frequency into plural signals, each of the plural signals being in a respective one of plural frequency ranges, the plural signals including a frequency range signal in one of the plural frequency ranges;

block length decision means, operating in response to an index, for determining a division of the frequency range signal into blocks to provide a block length decision signal indicating a block length for each of the blocks;

block floating processing means, operating in response to the block length decision signal from the block length decision means and in response to the index, for applying block floating processing to the blocks of the frequency range signal, each of the blocks having the block length indicated by the block length decision signal, the block floating processing circuit providing a block of a block floating processed frequency range signal from each of the blocks of the frequency range signal;

orthogonal transform means for orthogonally transforming the block of the block floating processed frequency range signal to produce plural spectral coefficients; and

adaptive bit allocation means for dividing the plural spectral coefficients from the orthogonal transform means into bands and for adaptively allocating a number of quantizing bits to quantize the spectral coefficients in each of the bands in response to an allowable noise level in each of the bands.

2. The apparatus of claim 1, wherein:

the frequency range signal includes plural words, each of the words having an absolute value; and

the apparatus additionally includes generating means for generating the index by calculating a logical sum of the absolute values of the words.

3. The apparatus of claims 1 or 2, wherein the orthogonal transform means includes a Discrete Cosine Transform circuit.

4. The apparatus of claim 1, wherein:

(a) the apparatus additionally comprises;

means for subdividing the frequency range signal into sub blocks, and

index generating means for generating an index for each of the sub blocks obtained by subdividing the frequency range signal;

(b) the block length decision means determines the division of the frequency range signal into blocks by comparing the indices of consecutive ones of the sub blocks of the frequency range signal to determine a number of the consecutive ones of the sub blocks to constitute each of the blocks; and

(c) the block floating processing means applies block floating processing to the blocks of the frequency range signal having the block length indicated by the block length decision signal using, for each one of the blocks, a block floating coefficient calculated from the indices of the sub blocks constituting the one of the blocks.

5. The apparatus of claim 4, wherein:

the frequency range signal includes plural words, each of the words having an absolute value; and

the index calculating means calculates the index for each one of the sub blocks by determining a maximum absolute value for the one of the sub blocks.

6. The apparatus of claim 4, wherein

the block length decision means includes:

(1) a comparing means for comparing the indices of the consecutive ones of the sub blocks to provide a comparison result, and

(2) a block defining means, responsive to the comparison result provided by the comparing means, for determining a division of the frequency range signal into blocks constituted a selected one of;

(i) one of the adjacent sub blocks;

(ii) two of the adjacent sub blocks; and

(iii) four of the adjacent sub blocks.

7. The apparatus of claim 6, wherein:

the means for subdividing the frequency range signal into sub blocks is additionally for subdividing the frequency range signal into frames, the frames including a frame constituted of a first sub block and a second sub block;

the comparing means is for comparing the indices of the first sub block and the second sub block to provide the comparison result; and

the block defining means determines a division of the frequency range signal wherein the frame of the frequency range signal constituted of the first sub block and the second sub block is divided into two equal blocks when the comparison result provided by the comparing means indicates that the index of the second sub block is twenty or more times the index of the first sub block, and otherwise determines a division of the frequency range signal wherein the frame is divided into a single block.

8. The apparatus of claim 6, wherein:

the means for subdividing the frequency range signal into sub blocks subdivides the frequency range signal into half sub blocks and into quarter sub blocks, and is additionally for subdividing the frequency range signal into frames, the frames including a frame constituted of a first half sub block and a second half sub block, the frame is also constituted of four quarter sub blocks, the four quarter sub blocks including pairs of consecutive ones of the quarter sub blocks, each of the pairs consisting of a first quarter sub block and a second quarter sub block;

the comparing means is for comparing the indices of the first half sub block and the second half sub block to provide a first comparison result, and is additionally for comparing the indices of the first quarter sub block and the second quarter sub block in each of the pairs of consecutive ones of the quarter sub blocks to provide a second comparison result; and

the block defining means determines a division of the frequency range signal wherein the frame of the frequency range signal is divided into;

four equal blocks when the second comparison result from the comparing means indicates that the index of the second quarter sub block of any one of the pairs of the consecutive ones of the quarter sub blocks is twenty or more times greater than the index of the first quarter sub block of the one of the pairs of the consecutive ones of the quarter sub blocks,

two equal blocks when the first comparison result from the comparing means indicates that the index of the second half sub block is ten or more times, but less than twenty times, the index of the first half sub block, and

a single block when the first comparison result from the comparing means indicates that the index of the second half sub block less than ten times the index of the first half sub block.

9. An apparatus for compressing a digital input signal, the apparatus comprising:

band division filter means for dividing the digital input signal into plural signals, each of the plural signals being in a respective one of plural frequency ranges, the plural signals including a frequency range signal in one of the plural frequency ranges;

block floating processing means for applying block floating processing to blocks of the frequency range signal to provide a block floating processed frequency range signal;

orthogonal transform means for orthogonally transforming blocks of the block floating processed frequency range signal to provide plural spectral coefficients; and

adaptive bit allocation means for dividing the spectral coefficients from the orthogonal transform means into bands and for adaptively allocating a number of quantizing bits for quantizing the spectral coefficients in each of the bands in response to an allowable noise level in each of the bands, the adaptive bit allocation means including;

allowable noise level calculation means for calculating the allowable noise level for each of the bands,

comparison means for comparing, in each of the bands, the allowable noise level with a minimum audible level and, for each of the bands in which the minimum audible level is higher than the allowable noise level, for setting a flag, and

means for selecting, in each of the bands in which the flag is set, the minimum audible level as the allowable noise level.

10. The apparatus of claim 9, wherein:

the adaptive bit allocation means quantizes the spectral coefficients using an actual number of quantizing bits;

the apparatus additionally includes;

means for providing an output signal including a target number of bits, and

means for determining an error between the actual number of bits and the target number of bits; and

the allowable noise level calculation means calculates the allowable noise level from an energy in each of the bands, and includes means for adjusting the allowable noise level in response to the error between the actual number of bits and the target number of bits.

11. The apparatus of claim 10, wherein the adaptive bit allocation means additionally includes means for adjusting the number of quantizing bits allocated to each of the bands by changing the allowable noise level.

12. The apparatus of claim 9, wherein:

the apparatus additionally comprises a block length decision means, operating in response to an index, for determining a division of the frequency range signal into the blocks;

the block floating processing means applies block floating processing to the blocks of the frequency range signal using the index as a block floating coefficient; and

the orthogonal transform means transforms the block floating processed frequency range signal divided into blocks determined by the block length decision means.

13. The apparatus of claims 9, 10, 11, or 12, wherein the orthogonal transform means includes a Discrete Cosine Transform (DCT) circuit.

14. The apparatus of claim 9, wherein:

the adaptive bit allocation means is additionally for dividing the spectral coefficients in one of the bands from the orthogonal transform means into plural sub bands, the plural sub bands including a lowest-frequency sub band; and

the comparison means is for comparing, in the one of the bands, the allowable noise level for the one of the bands with the minimum audible level for the lowest-frequency sub band, and for setting the flag for the one of the bands when the minimum audible level for the lowest-frequency sub band is higher than the allowable noise level.

15. An apparatus for compressing a digital input signal, the apparatus comprising:

index generating means for generating an index in response to the digital input signal;

block length decision means for determining a division of the digital input signal into blocks in response to the index;

block floating processing means for applying block floating processing to the blocks of the digital input signal in response to the index to provide block floating processed blocks of the digital input signal;

orthogonal transform means for orthogonally transforming each of the block floating processed blocks of the digital input signal to produce plural spectral coefficients; and

adaptive bit allocation means for dividing the plural spectral coefficients from the orthogonal transform means into bands, and for adaptively allocating a number of quantizing bits to quantize the spectral coefficients in each of the bands.

16. The apparatus of claim 15, wherein:

the digital input signal comprises plural words, each of the plural words having an absolute value; and

the index generating means generates the index by calculating a logical sum of the absolute values of the words.

17. The apparatus of claims 15 or 16, wherein the orthogonal transform means includes a Discrete Cosine Transform circuit.

18. The apparatus of claim 15, wherein:

(a) the index generating means generates an index for each of plural sub blocks obtained by dividing the digital input signal;

(b) the block length decision means includes comparing means for comparing the indices of consecutive ones of the sub blocks of the digital input signal to determine a number of the consecutive ones of the sub blocks of the digital input signal to constitute each one of the blocks; and

(c) the block floating processing means applies block floating processing to the blocks of the digital input signal determined by the block length decision means using, for each one of the blocks, a block floating coefficient calculated from the indices of the sub blocks constituting the one of the blocks.

19. The apparatus of claim 18, wherein:

the digital input signal includes plural words, each of the words having an absolute value; and

the index calculating means calculates the index for each one of the sub blocks by determining a maximum of the absolute values of the words in the one of the sub blocks.

20. The apparatus of claim 18, wherein:

(a) the index generating means includes;

(1) means for dividing the digital input signal into the sub blocks, and

(2) index calculating means for calculating an index for each of the sub blocks; and

(b) the block length decision means additionally includes block defining means, responsive to the comparing means, for determining a division of the digital input signal into blocks constituted of a selected one of;

(i) one of the sub blocks,

(ii) two consecutive ones of the sub blocks, and

(iii) four consecutive ones of the sub blocks.

21. The apparatus of claim 20, wherein:

the means for dividing the digital input signal into the sub blocks includes;

means for dividing the digital input signal into frames, and

means for dividing each of the frames into a first sub block and a second sub block;

the comparing means is for comparing the indices of the first sub block and the second sub block to provide a comparison result; and

the block defining means determines a division of the digital input signal into blocks in which the frames of the digital input signal for which the comparison result indicates that the index of the second sub block is twenty or more times the index of the first sub block are divided into two equal blocks, and otherwise determines a division of the digital input signal into blocks in which the frames are each divided into a single block.

22. The apparatus of claim 20, wherein:

(a) the means for dividing the digital input signal into sub blocks includes;

(1) means for dividing the digital input signal into flames, and

(2) means for dividing each of the flames into a first half sub block, a second half sub block, and into four quarter sub blocks, the four quarter sub blocks including pairs of consecutive ones of the quarter sub blocks, each of the pairs consisting of a first quarter sub block and a second quarter sub block;

(b) the comparing means is for comparing the indices of the first half sub block and the second half sub block to provide a first comparison result, and for comparing the indices of pairs of the first quarter sub block and the second quarter sub block in each of the pairs of consecutive ones of the quarter sub blocks to provide a second comparison result; and

(c) the block defining means determines a division of the digital input signal into blocks in which;

(1) ones of the flames for which the second comparison result indicates that the index of the second quarter sub block of any one of the pairs of consecutive ones of the quarter sub blocks is twenty or more times greater than the index of the first of the one of the pairs of the consecutive ones of the quarter sub blocks are divided into four equal blocks,

(2) ones of the flames for which the first comparison result indicates that the index of the second half sub block is ten or more times, but less than twenty times, the index of the first half sub block is divided into two equal blocks, and

(3) ones of the flames for which the first comparison result indicates that the index of the second half sub block is less than ten times the index of the first half sub block are divided into a single block.

23. The apparatus of claim 15, wherein the adaptive bit allocation means is for dividing the spectral coefficients into bands corresponding to critical bands.

24. The apparatus of claim 15, wherein the adaptive bit allocation means is for dividing the spectral coefficients towards higher frequencies into bands corresponding to a fraction of a critical band.

25. An apparatus for compressing a digital input signal, the apparatus comprising:

block length decision means for determining a division of the digital input signal into blocks in response to an index;

block floating means for applying block floating processing to each of the blocks of the digital input signal using the index as a block floating coefficient; and

means for deriving spectral coefficients from the block floating processed blocks of the digital input signal; and

adaptive bit allocation means for dividing the spectral coefficients by frequency into bands and for adaptively allocating a number of quantizing bits for quantizing the spectral coefficients in each of the bands in response to an allowable noise level for each of the bands, the adaptive bit allocation means including;

allowable noise level calculation means for calculating an allowable noise level for each of the bands,

comparing means for comparing, in each of the bands, the allowable noise level with a minimum audible level, and

selecting means for selecting the minimum audible level as the allowable noise level in each of the bands for which the comparing means determines that the minimum audible level is higher than the allowable noise level.

26. The apparatus of claim 25, wherein the means for deriving spectral coefficients from the digital input signal includes an orthogonal transform circuit.

27. The apparatus of claim 26, wherein the orthogonal transform circuit is a Discrete Cosine Transform (DCT) circuit.

28. The apparatus of claims 25 or 26, wherein:

the bands into which the adaptive bit allocation means divides the spectral coefficients include a band corresponding to a critical band;

the adaptive bit allocation means is additionally for dividing the spectral coefficients in the band corresponding to a critical band into sub bands, the sub bands including a lowest-frequency sub band;

the comparing means is for comparing, in the band corresponding to a critical band, the allowable noise level for the band corresponding to a critical band with the minimum audible level for the lowest-frequency sub band; and

the selecting means is for selecting, as the allowable noise level for the sub bands in the band corresponding to a critical band, the respective minimum audible levels for the sub bands when the comparing means indicates that the minimum audible level for the lowest-frequency sub band is higher than the allowable noise level.

29. The apparatus of claims 25 or 26, wherein:

the number of quantizing bits adaptively allocated by the adaptive bit allocation means is an actual number of bits;

the apparatus additionally includes;

means for providing an output signal including a target number of bits, and

means for determining an error between the actual number of bits and the target number of bits; and

the allowable noise level calculation means includes means for adjusting the allowable noise level in ones of the bands in response to the error between the actual number of bits and the target number of bits.

30. The apparatus of claim 29, wherein the adaptive bit allocation means adjusts the number of quantizing bits allocated to the bands in response to changes in the allowable noise level in the ones of the bands caused by the means for adjusting the allowable noise level.

31. A method for compressing a digital input signal, the method comprising the steps of:

generating an index in response to the digital input signal;

determining a division of the digital input signal into blocks in response to the index;

applying block floating processing to the blocks of the digital input signal in response to the index to provide block floating processed blocks of the digital input signal;

orthogonally transforming each of the block floating processed blocks of the digital input signal to produce plural spectral coefficients; and

dividing the plural spectral coefficients into bands, and adaptively allocating a number of quantizing bits to quantize the spectral coefficients in each of the bands.

32. The method of claim 31, wherein:

the digital input signal comprises plural words, each of the plural words having an absolute value; and

in the step of generating an index, the index is generated by calculating a logical sum of the absolute values of the words.

33. The method of claims 31 or 32, wherein, in the step of orthogonally transforming each of the block floating processed blocks of the digital input signal, each of the block floating processed blocks of the digital input signal is orthogonally transformed using a discrete cosine transform.

34. The method of claim 31, wherein:

(a) in the step of generating an index, an index is generated for each of plural sub blocks obtained by dividing the digital input signal;

(b) the step of determining a division of the digital input signal into blocks includes a step of comparing the indices of consecutive sub blocks of the digital input signal to determine a number of the consecutive sub blocks of the digital input signal to constitute each one of the blocks; and

(c) in the step of applying block floating processing to the blocks of the digital input signal, block floating processing is applied to the blocks of the digital input signal using, for each one of the blocks, a block floating coefficient calculated from the indices of the sub blocks constituting the one of the blocks.

35. The method of claim 34, wherein:

the digital input signal includes plural words, each word having an absolute value; and

in the step of generating an index, the index for each one of the sub blocks is calculated by determining a maximum of the absolute values of the words in the one of the sub blocks.

36. The method of claim 34, wherein:

(a) the step of generating an index includes steps of;

(1) dividing the digital input signal into sub blocks, and

(2) calculating an index for each of the sub blocks; and

(b) in the step of determining a division of the digital input signal into blocks, a division of the digital input signal into blocks constituted of a selected one of one sub block, two sub blocks, and four sub blocks is determined in response to the step of comparing the indices of consecutive sub blocks.

37. The method of claim 36, wherein:

(a) the step of dividing the digital input signal into sub blocks includes steps of;

(1) dividing the digital input signal into frames, and

(2) dividing each frame into a first sub block and a second sub block;

(b) in the step of the comparing the indices of consecutive sub blocks, the indices of the first sub block and the second sub block are compared; and

(c) in the step of determining a division of the digital input signal into blocks, a division of each frame of the digital input signal into two equal blocks is determined when the step of comparing determines that the index of the second sub block is twenty or more times the index of the first sub block, and a division of the frame of the digital input signal into a single block is otherwise determined.

38. The method of claim 36, wherein:

(a) the step of dividing the digital input signal into sub blocks includes steps of;

(1) dividing the digital input signal into frames, and

(2) dividing each frame into a first half sub block, a second half sub block, and into our quarter sub blocks;

(b) in the step of comparing the indices of consecutive sub blocks, the indices of the first half sub block and the second half sub block are compared, and the indices of pairs of consecutive quarter sub blocks are compared, each pair of consecutive quarter sub blocks including a first quarter sub block and a second quarter sub block; and

(c) in the step of determining a division of the digital input signal into blocks, the division of each frame of the digital input signal is determined as follows;

(1) into four equal blocks when the step of comparing indicates that the index of the second of any pair of consecutive quarter sub blocks is twenty or more times greater than the index of the first of any pair of consecutive quarter sub blocks,

(2) into two equal blocks when the step of comparing indicates that the index of the second half sub block is ten or more times, but less than twenty times, the index of the first half sub block, and

(3) into a single block when the step of comparing indicates that the index of the second half sub block less than ten times the index of the first half sub block.

39. The method of claim 31, wherein, in the step of dividing the plural spectral coefficients into bands, the spectral coefficients are divided into bands corresponding to critical bands.

40. The method of claim 31, wherein, in the step of dividing the plural spectral coefficients into bands, the spectral coefficients towards higher frequencies are divided into bands corresponding to a fraction of a critical band.

41. The method of claim 31, wherein:

the method additionally comprises a step of dividing the digital input signal into plural signals, each of the plural signals being in a respective one of plural frequency ranges, the plural signals including a frequency range signal in one of the plural frequency ranges;

in the step of generating an index in response to the digital input signal, an index is generated in response to the frequency range signal;

in the step of determining a division of the digital input signal into blocks, the division of the frequency range signal into blocks is determined in response to the index;

in the step of applying block floating processing to the digital input signal, block floating processing is applied to the blocks of the frequency range signal in response to the index to provide block floating processed blocks of the frequency range signal; and

in the step of orthogonally transforming each of the block floating processed blocks of the digital input signal, each of the block floating processed blocks of the frequency range signal is orthogonally transformed to produce ones of the plural spectral coefficients.

42. A method for compressing a digital input signal, the method comprising steps of:

determining a division of the digital input signal into blocks in response to an index;

applying block floating processing to each of the blocks of the digital input signal using the index as a block floating coefficient;

deriving spectral coefficients from the block floating processed blocks of the digital input signal;

dividing the spectral coefficients by frequency into bands; and

adaptively allocating a number of quantizing bits for quantizing the spectral coefficients in each of the bands in response to an allowable noise level for each of the bands, the step of adaptively allocating a number of quantizing bits including steps of;

calculating an allowable noise level for each of the bands,

comparing, in each of the bands, the allowable noise level with a minimum audible level, and

selecting the minimum audible level as the allowable noise level in each of the bands for which the step of comparing determines that the minimum audible level is higher than the allowable noise level.

43. The method of claim 42, wherein the step of deriving spectral coefficients from the digital input signal includes a step of orthogonally transforming the digital input signal.

44. The method of claim 43, wherein, in the step of orthogonally transforming the digital input signal, the digital input signal is orthogonally transformed using a discrete cosine transform.

45. The method of claims 42 or 43, wherein the method is for compressing the digital input signal to provide a compressed signal including a target number of bits, and wherein:

in the step of adaptively allocating a number of quantizing bits, the number of bits adaptively allocated is an actual number of bits;

the method additionally includes a step of determining an error between the actual number of bits and the target number of bits; and

the step of adaptively allocating a number of quantizing bits includes a step of adjusting the allowable noise level in ones of the bands in response to the error between the actual number of bits and the target number of bits.

46. The method of claim 45, wherein, in the step of adaptively allocating a number of quantizing bits, the number of quantizing bits allocated to the ones of the bands and the sub bands is adjusted in response to changes in the allowable noise level caused by the step of adjusting the allowable noise level.

47. The method of claims 42 or 43, wherein:

in the step dividing the spectral coefficients into bands, the spectral coefficients are divided into bands including a band corresponding to a critical band, and the spectral coefficients in the band corresponding to a critical band are divided into sub bands, the sub bands including a lowest-frequency sub band;

in the step of comparing, the allowable noise level for the band corresponding to a critical band is compared with the minimum audible level for the lowest-frequency sub band; and

in the step of selecting, the minimum audible level for the lowest-frequency sub band is selected as the allowable noise level for the band corresponding to a critical band when the step of comparing indicates that the minimum audible level for the lowest-frequency sub band is higher than the allowable noise level.

48. The method of claim 42, wherein:

the method additionally comprises a step of dividing the digital input signal into plural signals, each of the plural signals being in a respective one of plural frequency ranges, the plural signals including a frequency range signal in one of the plural frequency ranges;

the step of determining a division of the digital input signal into blocks includes a step of generating the index in response to the frequency range signal;

in the step of determining a division of the digital input signal into blocks, a division of the frequency range signal into blocks is determined in response to the index;

in the step of applying block floating processing to each of the blocks of the digital input signal, block floating is applied to the blocks of the frequency range signal in response to the index to generate a block of a block floating processed frequency range signal from each of the blocks of the frequency range signal; and

the step of deriving spectral coefficients from the block floating processed blocks of the digital input signal includes a step of orthogonally transforming the block floating processed block of the frequency range signal to produce ones of the spectral coefficients.

49. An apparatus for expanding a compressed digital signal including plural quantized spectral coefficients and auxiliary information, the apparatus comprising:

adaptive bit allocation decoding means, operating in response to the auxiliary information, for inversely quantizing the quantized spectral coefficients to provide plural spectral coefficients;

block floating means for applying inverse block floating to the spectral coefficients to provide inverse block floating processed spectral coefficients;

inverse orthogonal transform means for inversely orthogonally transforming the inverse block floating processed spectral coefficients to provide plural frequency range signals; and

inverse filter means for synthesizing the frequency range signals to provide an output signal.

50. The apparatus of claim 49, wherein the inverse orthogonal transform means includes an inverse discrete cosine transform circuit.

51. The apparatus of claim 49, wherein:

(a) the apparatus is for expanding a compressed digital signal wherein;

(1) the spectral coefficients are quantized in critical bands,

(2) the critical bands include a divided band, the divided band being a higher-frequency one of the critical bands that is divided into plural sub bands, the sub bands including a lowest-frequency sub band, and

(3) the auxiliary information includes an allowable noise level for each of the critical bands, the allowable noise level for the divided band being the allowable noise level for the lowest-frequency sub band; and

(b) the apparatus additionally comprises means for determining an allowable noise level for each sub band of the divided band in response to the allowable noise level for the divided band.

52. A method for expanding a compressed digital signal to provide a digital output signal, the compressed digital signal including:

(a) plural quantized spectral coefficients divided by frequency into bands, the bands including a divided band in which the spectral coefficients therein are further divided by frequency into sub bands, the quantized spectral coefficients in each of the bands and each of the sub bands being quantized using an adaptively-allocated number of quantizing bits,

(b) an allowable noise level for each band, and

(c) a flag signal for the divided band,

the method comprising the steps of;

setting the allowable noise level of the divided band as the allowable noise level for the divided band when the flag signal for the divided band is in a first state, and setting the allowable noise level of the divided band as the allowable noise level for one of the sub bands constituting the divided band when the flag signal for the divided band is in a second state;

determining, when the flag signal for the divided band is second state, from the allowable noise level of the divided band, an allowable noise level for each of the other ones of the sub bands constituting the divided band;

using the allowable noise level for each of the bands and for each of the sub bands constituting the divided band to inversely quantize the respective quantized spectral coefficients in each of the bands and in each of the sub bands constituting the divided band to provide spectral coefficients; and

deriving the digital output signal from the spectral coefficients.

53. The method of claim 52, wherein:

when the flag signal for the divided band is in the second state, the allowable noise level for the divided band is the allowable noise level for the lowest-frequency one of the sub bands constituting the divided band; and

in the step of determining an allowable noise level for each of other ones of the sub bands constituting the divided band, the allowable noise level for each of the other ones of the sub bands higher in frequency than the lowest-frequency one of the sub band is calculated.

54. The method of claim 52, wherein:

the method additionally includes a step of providing a read-only memory wherein allowable noise levels are stored; and

the step of determining an allowable noise level for each of other ones of the sub bands constituting the divided band includes a step of reading an allowable noise level for each of the other ones of the sub bands from the read-only memory in response to the allowable noise level for the divided band.

55. The method of claim 52, wherein the step of deriving the digital output signal from the spectral coefficients includes steps of:

dividing the spectral coefficients by frequency into plural frequency ranges;

inversely orthogonally transforming the spectral coefficients in each of the frequency ranges to provide frequency range signals; and

synthesizing the frequency range signals to provide the digital output signal.

56. The method of claim 52, wherein, in the compressed digital signal, the plural quantized spectral coefficients are divided by frequency into bands corresponding to critical bands.

57. An apparatus for compressing a digital input signal, the apparatus comprising:

means for deriving spectral coefficients from the digital input signal;

frequency dividing means for dividing the spectral coefficients by frequency into bands, the bands including a band corresponding to a critical band, and additionally for subdividing the spectral coefficients in the band by frequency into sub bands, the sub bands including a lowest-frequency sub band;

allowable noise level calculation means for calculating a allowable noise level for each of the bands;

supplying means for supplying a minimum audible level for each of the bands except the band, and for each of the sub bands in the band;

comparing means for determining, in each of the bands except the band, when the minimum audible level supplied by the supplying means is greater than the allowable noise level calculated by the allowable noise level calculation means, and for determining, in the band, when the minimum audible level supplied by the supplying means for only the lowest-frequency sub band is greater than the allowable noise level calculated by the allowable noise level calculation means for the band;

substituting means for substituting, in each one of the bands in which the comparing means determines that the minimum audible level is greater than the allowable noise level, the minimum audible level supplied by the supplying means for the allowable noise level calculated by the allowable noise level calculating means as the allowable noise level for the one of the bands, and for substituting, in each one of the sub bands in the band when the comparing means determines that the minimum audible level for the lowest-frequency sub band is greater than the allowable noise level for the band, the minimum audible level supplied for the one of the sub bands by the supplying means for the allowable noise level calculated for the band by the allowable noise level calculation means as the allowable noise level for the one of the sub bands; and

adaptive bit allocation means for adaptively allocating a number of quantizing bits among the bands and the sub bands for quantizing the spectral coefficients therein, the bit allocation means allocating quantizing bits among the bands and the sub bands in response to the allowable noise level for each of the bands and sub bands.

58. The apparatus of claim 57, wherein the means for deriving spectral coefficients from the digital input signal includes an orthogonal transform circuit.

59. The apparatus of claim 58, wherein the orthogonal transform circuit is a Discrete Cosine Transform (DCT) circuit.

60. The apparatus of claim 57, wherein:

the number of quantizing bits adaptively allocated by the adaptive bit allocation means is an actual number of bits;

the apparatus additionally includes;

means for providing an output signal including a target number of bits, and

means for determining an error between the actual number of bits and the target number of bits; and

the allowable noise level calculation means includes means for adjusting the allowable noise level in ones of the bands and the sub bands in response to the error between the actual number of bits and the target number of bits.

61. The apparatus of claim 60, wherein the adaptive bit allocation means adjusts the number of quantizing bits adaptively allocated to the bands in response to changes in the allowable noise level in the ones of the bands and the sub bands caused by the means for adjusting the allowable noise level.

62. The apparatus of claim 57, wherein the means for deriving spectral coefficients from the digital input signal includes:

block length decision means for determining a division of the digital input signal into blocks in response to an index;

block floating means for applying block floating processing to each block of the digital input signal using the index as a block floating coefficient; and

means for deriving the spectral coefficients from the block floating processed blocks of the digital input signal.

63. A method for compressing a digital input signal, the method comprising steps of:

deriving spectral coefficients from the digital input signal;

dividing the spectral coefficients by frequency into bands, the bands including a band corresponding to a critical band;

subdividing the spectral coefficients in the band by frequency into sub bands, the sub bands including a lowest-frequency sub band;

calculating an allowable noise level for each of the bands;

supplying a minimum audible level for each of the bands except the band, and for each of the sub bands in the band;

determining, in each of the bands except the band, when the minimum audible level supplied in the supplying step is greater than the allowable noise level calculated in the allowable noise level calculating step, and determining, in the band, when the minimum audible level supplied in the supplying step for only the lowest-frequency sub band is greater than the allowable noise level calculated for the band in the allowable noise level calculating step;

substituting, in each one of the bands in which determining step determines that the minimum audible level is greater than the allowable noise level, the minimum audible level supplied in the supplying step for the allowable noise level calculated in the allowable noise level calculating step as the allowable noise level for the one of the bands, and substituting, in each one of the sub bands in the band when the determining step determines that the minimum audible level for the lowest-frequency sub band is greater than the allowable noise level for the band, the minimum audible level supplied for the one of the sub bands in the supplying step for the allowable noise level calculated for the band in the allowable noise level calculating step as the allowable noise level for the one of the sub bands; and

adaptively allocating a number of quantizing bits among the bands and the sub bands for quantizing the spectral coefficients therein, the quantizing bits being adaptively allocated to each of the bands and the sub bands in response to the allowable noise level for each of the bands and sub bands.

64. The method of claim 63, wherein the step of deriving spectral coefficients from the digital input signal includes a step of orthogonally transforming the digital input signal.

65. The method of claim 64, wherein, in the step of orthogonally transforming the digital input signal, the digital input signal is orthogonally transformed using a discrete cosine transform.

66. The method of claim 63, wherein the method is for compressing the digital input signal to provide a compressed signal including a target number of bits, and wherein:

in the step of adaptively allocating a number of quantizing bits, the number of bits adaptively allocated is an actual number of bits;

the method additionally includes a step of determining an error between the actual number of bits and the target number of bits; and

the step of adaptively allocating a number of quantizing bits includes a step of adjusting the allowable noise level in ones of the bands and the sub bands in response to the error between the actual number of bits and the target number of bits.

67. The method of claim 66, wherein, in the step of adaptively allocating a number of quantizing bits, the number of quantizing bits allocated to each of the bands and sub bands is adjusted in response to changes in the allowable noise level in the ones of the bands and the sub bands caused by the step of adjusting the allowable noise level.

68. The method of claim 63, wherein the step of deriving spectral coefficients from the digital input signal includes steps of:

determining a division of the digital input signal into blocks in response to an index;

applying block floating processing to each block of the digital input signal using the index as a block floating coefficient; and

deriving the spectral coefficients from the block floating processed blocks of the digital input signal.